Title: Forward masking of auditory nerve (N1) and brainstem responses (Wave V) in humans
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00102819/00001
 Material Information
Title: Forward masking of auditory nerve (N1) and brainstem responses (Wave V) in humans
Physical Description: Book
Language: English
Creator: Kramer, Steven John, 1950-
Copyright Date: 1981
 Record Information
Bibliographic ID: UF00102819
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: oclc - 07891046
ltuf - ABS1781

Full Text







I would like to thank the chairman of my supervisory

committee, Dr. Donald C. Teas, for all his tireless efforts

and encouragements in helping me to conceive, organize,

and carry out this research. I would also like to thank

my committee members, Dr. W. Keith Berg, Dr. William E.

Brownell, Dr. F. Joseph Kemker, and Dr. Kenneth J. Gerhardt,

for their support and participation. In addition I would

like to express thanks to my friends and subjects without

whom this research would not have been possible and to

Mrs. Kathy Farley for her secretarial assistance. Finally,

I would like to express my appreciation to Kathleen Cannon

for all her help, encouragement and sacrifice and to whom

this project is dedicated.






. vii

. xi




General Description of the AP. .
Forward Masking of the Whole-
Nerve Action Potential (AP) .
General Description of the ABR
Forward Masking of the Auditory
Brainstem Response (ABR) .

II METHODS . . . . . . .

Subjects . . . . . .
Equipment . . . . .

Stimuli . . . . . .
Electrodes . . . .
Procedures . . . .

II RESULTS . . . . . . .

"Standard" Recovery Function .
The Effect of Masker Duration..
The Effect of Masker Level . .
The Effect of High-Pass Masking.
Derived Response . . . .

IV DISCUSSION . . . . . .

. 14

S 17
. 28

. 32

S. 47

. 47
S. 48

. 48
. 54
S. 55

S. 58

S. 65
. 81
. 91
S. 99
. 119

S. 129

. ii



General Effects of Forward
Masking . . . . . ... 132
Effect of Masker Duration and
Level . . . . . . . 134
Effects of Masker Low Frequency
Cutoff (High-Pass Masking) . .138
Possible Mechanism Underlying the
Effects of Forward Masking on
N1 and Wave V . . . . 141

Wave V Amplitude . . ... 142
V-N1 vs N1 Amplitude. ... . 144
Central Physiological Evidence. 148
Explanation of Forward
Masking Effects on N1 .. 150

Relation to Psychophysics. .... 153
Summary and Conclusions. ... . 154

BIBLIOGRAPHY. . . . . . . . .. 157

BIOGRAPHICAL SKETCH . . . . . ... .172


Table Page

I Forward masking series received by
each of the subjects. (+) indicates
success and (-) indicates failure
in obtaining data for ABR/AP. . .. 60

II Means and standard deviations
(S.D.) of wave V and N1 latencies
for the "standard" series . . . 72

III Means and standard deviations (S.D.)
of wave V and N1 amplitudes for the
"standard" series . . . ... .79

IV Means and standard deviations (S.D.)
of wave V, N1 and V-N1 probe response
latencies for the three masker
durations . . . . . . .. 83

V Means and standard deviations (S.D.)
of wave V and N1 probe response
amplitudes for the three masker
durations . . . . . . .. 89

VI Means and standard deviations of
wave V, N1 and V-N1, probe response
latencies for the three masker
levels. . . . . . . . . 94

VII Means and standard deviations of
wave V and N1 probe response
amplitudes for the three masker
levels. . . . . . . . ... 97

LIST OF TABLES (Continued)

Table Page

VIII Means and standard deviations (S.D.)
of wave V, N1, and V-N1 probe
response latencies for the four
masker low frequency cutoffs. . . ... 110

IX Means and standard deviations (S.D.)
of wave V and N1 probe response
amplitudes for the four masker low
frequency cutoffs . . . . .. 118

X Summary of significant effects and
interactions of masker parameters
on wave V and N1 probe responses. ... .135


Figure Page

1 Spectra of the masking stimuli (a)
and frequency response of the TDH-39
earphone (b) . . . . . . .50

2 Simplified block diagram of the
stimulus generating and response
recording equipment . . . ... .52

3 Examples of unmasked control response
waveforms . . . . . . . 62

4 Example of ABR and AP probe response
waveforms as a function of AT for the
"standard" series . . . . ... .68

5 Mean and fl S.D. of wave V, N1, and
V-N1 latencies as a function of AT
(delay time) for the "standard"
series. . . . . . . . ... 69

6 Mean and 1 S.D. for wave V and N1
amplitude (uvolts) as a function of
AT for the "standard" series . . 76

7 Mean wave V and N1 amplitude recovery
functions expressed as percent of the
mean unmasked control amplitude (100%)
for the "standard" series . . .. 77

8 Summary comparison of wave V and N
for the "standard" series .. ... 80

9 Latency recovery functions for wave V,
N1 and V-N1 probe responses following
wide-band makers with durations of
30, 60, and 120 msec. . . . ... .86



Figure rage

10 Amplitude recovery functions for
wave V probe responses following
wide-band makers with durations
of 30, 60, and 100 msec . . ... .87

11 Amplitude recovery functions for
N1 probe responses following wide-
band makers with durations of 30,
60, and 120 msec. . . . . ... .90

12 Latency recovery functions for
wave V, N1, and V-N, probe responses
following 60 msec wide-band makers
with levels of +10 dB relative to
the "standard" series (referred to
as 0 dB). . . . . . ... .. .. 93

13 Amplitude recovery functions for
wave V probe responses following
60 msec wide-band makers with
levels of 10 dB relative to the
"standard" series . . . . ... .96

14 Amplitude recovery functions for
N1 probe responses following 60
msec wide-band makers with levels
of 10 dB relative to the "standard"
series. . . . . . . . ... 98

15 Example of the probe ABRs as a
function of AT (top to bottom) for
the different masker low frequency
cutoffs (left to right) . . ... 102

16 Example of probe APs as a function
of AT (top to bottom) for the
different masker low frequency
cutoffs (left to right) . . ... 105




Latency recovery functions for
wave V, N1, and V-N1 probe re-
sponses following 60 msec makers
with masker low frequency cutoffs
of .02, 2, 4, and 6 kHz . . . .

Wave V and N1 probe response
latencies as a function of masker
low frequency cutoff . . . . .

Amplitude recovery functions for
wave V probe responses following
60 msec wide-band makers with
masker low frequency cutoffs of
.02, 2, 4, and 6 kHz. . . . . .

Wave V probe response amplitude as
a function of masker low frequency
cutoff . . . . . . . .

Amplitude recovery functions for
N1 probe responses following 60
msec wide-band makers with masker
low frequency cutoffs of .02, 2,
4, and 6 kHz . . . . . .

N1 probe response amplitude as a
function of masker low frequency
cutoff . . . . . . .

Example of derived response
technique for the ABR . . . .

Example of the derived response
technique for the AP. . . . .

Mean latencies of wave V and N1
as a function of derived CF . ..






. 117

. 122

. 124

. 125



Figure Page

26 Mean amplitudes of wave V and
N1 as a function of derived CF. ... . 127

27 V-N1 latency interval as a function
of % N1 control amplitude for the
different masker low frequency cutoffs
at each AT . . . . . . . 147

Abstract of Dissertation Presented to the Graduate Council
of the University of Florida in Partial Fulfillment of
the Requirements for the Degree of
Doctor of Philosophy



Steven John Kramer

June 1981

Chairman: Donald C. Teas
Major Department: Speech

In both psychophysical and electrophysiological inves-

tigations, it has been demonstrated that a response to one

sound (probe) can be influenced by another sound maskerr)

when the two sounds occur simultaneously or in a temporal

relation. The relation in which the masker precedes the

probe is called forward masking. Measurements of the probe

response as a function of the time interval following the

masker (AT) describe the forward masking recovery func-


In the present study, the effects of forward masking

were investigated on wave V of the ABR along with N1

responses recorded simultaneously (on a separate channel)

from the ear canal. Evoked potentials to 50 dB HL

broadband clicks were obtained for ATs of 6.2, 12.5, 25,

50, and 100 msec following short-duration noise makers.

Postmasking recovery functions were measured for different

masker durations, masker levels, and masker low frequency

cutoffs (high-pass masking).

The largest amount of masking occurred for the

shortest ATs and masking decreased as AT increased. The

effects on wave V were found to be different than on N1.

The primary effects of forward masking on the probe re-

sponses were decreases in N1 amplitude and increases in

wave V latency, neither of which were fully recovered by

AT=100 msec. Wave V amplitude recovered by AT=25 msec

and showed a robustness to the forward masking. Charac-

teristic changes in the recovery functions were found for

variations on masker parameters.

The relation between wave V latency and N1 amplitude,

along with other characteristics of the results, suggests

that underlying mechanisms include a recoding of the

neural input (N1) within the brainstem pathways generating

the wave V potential.



The effects that one sound may have on the perception of

another sound have been of interest to investigators ever since

Mayer (1876a,b) first described some aspects of this basic

phenomenon which has become known as masking. Masking is a

psychophysical term used to denote an elevation in threshold

of audibility for one sound (probe) in the presence of another

sound maskerr). Mayer observed that a watch-tick could be

obliterated by a louder clock-tick when the two ticks over-

lapped each other. By using two time-pieces running at

different speeds, Mayer also noticed that the quieter watch-

tick was obliterated by the clock-tick at instances when the

two ticks did not exactly overlap in time. Mayer's early

qualitative observations demonstrated that different temporal

relations among sounds could influence the perception of those

sounds and have since led to an abundance of research known,

in a general sense, as temporal masking. The masking situation

with the masker coincident with the probe is generally called

simultaneous masking. The term forward masking is used for

situations where the masker precedes the probe and backward

masking applies to situations where the masker follows the

probe. The time interval between the masker and the probe

is generally called AT. One problem with the simultaneous

masking paradigm is that nonlinear distortion products are

produced by the two stimuli with different frequencies, thereby

producing energy at frequencies other than the stimulus fre-

quencies (Rodenburg, Verschuure, and Brocaar, 1974; Shannon,

1975; Wightman, McGee, and Kramer, 1977; Moore, 1978; Vogten,

1978). A forward masking paradigm overcomes the distortion

problem since the two stimuli are not present at the same time.

Forward and backward masking paradigms have been used in

psychophysical investigations to describe the time course of

masking effects (Lusher and Zwislocki, 1947; Zwislocki, Pirodda,

and Rubin, 1959; Elliott, 1962; Plomp, 1964; Wilson and Carhart,

1971; Robinson and Pollack, 1973; Duifhuis,1973; Penner, 1974;

Fastl, 1976; Berg and Yost, 1976). In this procedure, the

threshold of the probe stimulus is measured as a function of

AT. As AT increases, less energy is required to detect

the probe signal and the plot of threshold vs At defines the

masking recovery function for stated masker-probe parameters.

In general, the decrement in sensitivity to the probe produced

by the forward or backward masker is greatest for shortest ATs

(least sensitivity) and then monotonically decreases (toward

greater sensitivity) as AT increases. The decrease in

sensitivity to the probe becomes greater for longer masker

durations and higher masker levels. The actual descriptions

of the recovery functions are not entirely agreed upon and

are dependent on stimulus parameters. Recovery from backward

masking occurs much faster than recovery from forward masking.

Backward masking will not be addressed any further in this


In electrophysiological investigations, masking paradigms

have been adapted to provide correlatives to psychophysical

masking phenomena. The investigation of electrical potentials

evoked by acoustic stimuli has been of interest to auditory

researchers as a means of describing underlying physiological

phenomena that may possibly mediate some of the psychophysical

observations. The development of signal-averaging computers

in the 1960's greatly improved the ability to extract stimulus-

related electrical potentials from the random electrical back-

ground activity and has made it possible to record electrical

activity to auditory stimuli from remote sites within the

boundaries of the skull and scalp. Two such averaged evoked

potentials most applicable to humans are the whole-nerve

action potential (AP) and the Auditory Brainstem Response

(ABR), which are the subjects of the present study.

The term masking has been retained in electrophysiologi-

cal experiments and refers to a reduction in some measure of

neural response to one signal (probe) due to the presence of

another signal maskerr). In physiological masking, it has

been thought that the masker "subtracts" the contributions

(through refractory, adaptation, and/or desynchronizing

mechanisms) from certain areas of the cochlear partition

dependent on the level and spectral characteristics of the

masker. A response to the probe, recorded in the presence

of the masker, is considered to reflect the "unmasked" portion

of the cochlear partition (Teas, Eldredge, and Davis, 1962;

Elberling, 1974; Spoor, Eggermont and Odenthal, 1976; Don and

Eggermont, 1978; Parker and Thornton, 1978). The probe and

masker can be presented simultaneously or in a temporal

relation, e.g., forward masking.

In both psychophysical and electrophysiological masking

studies, there is general agreement that for relatively short

duration makers at levels below 80-90 dB Sound Pressure Level

(SPL) the decrement of probe response activity is due to some

short-term noncumulative effect. For makers above 80-90 dB

SPL the effects are different and involve some form of cumula-

tive "fatigue" (Hood, 1950; Gisselsson and Sorensen, 1959;

Coats, 1964a; Smith, 1977). This investigation is concerned

only with the short-term lower level effects of forward


Forward masking effects have been investigated by several

researchers in recordings of the whole-nerve response (AP)

(Hawkins and Kniazuk, 1950; Sorensen, 1959; Coats, 1964a,b,

1967, 1971; Spoor, 1965; Coats and Dickey, 1972; Eggermont

and Spoor, 1973a,b; Eggermont and Odenthal, 1974a; Spoor et al.,

1976; Charlet de Sauvage and Aran, 1976; Dallos and Cheatham,

1976; Harris, 1979; Huang and Buchwald, 1980; Abbas and Gorga,

1981). In general, these studies have shown that the amounts

of post masking recovery functions are dependent on the pro-

cedure, masker and probe levels, and masker durations. A re-

view of existing literature on forward masking of the AP is

presented in more detail in a later section.

A special case of forward masking is that of repetitive

stimuli. Responses to successive stimuli show a decrement

when the preceding stimuli have an extended effect. The re-

sponse to each stimulus can be measured for a pair of stimuli

or for a train of stimuli as a function of ISI. The amount

of response decrement for the AP has been shown to be greater

for shorter ISIs and the final amount of response decrement

is reached by the 4th or 5th stimulus (Sorensen, 1959; Kupper-

man, 1971; Yoshie and Ohashi, 1971; Eggermont and Spoor, 1973a;

Thornton and Coleman, 1975; Spoor et al., 1976; Don, Allen

and Starr, 1977). Since no further response decrement of

the AP occurs after the 5th stimulus in the stimulus-train

experiments, others have used repetition rate procedures

where the final averaged AP responses to fast repetition rates

are compared to responses using a slow repetition rate (Peake,

Goldstein, and Kiang, 1962; Yoshie, 1968; Teas and Henry, 1969;

Spoor et al., 1976). Response magnitude as a function of ISI

defines the recovery function in the repetitive stimuli para-


Forward masking paradigms have also been used to measure

masker and probe responses in single auditory nerve fibers

(Galambos and Davis, 1943; Kiang, Watanabe, Thomas, and Clark,

1965; Young and Sachs, 1973; Evans, 1976; Smith, 1977, 1979;

Bauer, 1978; Harris and Dallos, 1979; Abbas, 1979; Delgutte,

1980). These studies have shown that during continuous stimu-

lation (of constant intensity and frequency), e.g., by the

masker, the discharge rate of the nerve fiber responds

maximally at the onset then declines rapidly at first followed

by a more gradual decline until a quasi steady-state is reached

by 100-200 msec. This (normal) neural decrement is called

adaptation. Following the offset of a masker to which the

single fiber has adapted, the spontaneous neural activity

shows an immediate depression followed by a recovery to base-

line levels. The probe response following the masker offset

shows a decrement in discharge rate compared to its unmasked

value which then recovers as AT is lengthened. The recovery

of the probe response magnitude has been shown to be related

to the recovery of the spontaneous activity and this is pro-

portional to the driven onset (and average) discharge rate

which occurs for the masker. Both the adaptation and recovery

from adaptation have been shown to follow exponential recovery

functions; however, the time course of adaptation is much

faster than the time course of recovery from adaptation

(Harris and Dallos, 1979). In addition, the recovery func-

tions have been shown to have different time constants for

different species, e.g., they are about 2.5 times longer in

guinea pig than chinchilla. Single auditory nerve fiber

responses to increasing click repetition rate show decreases

in neural discharge rate and only small (0.1 msec) latency

increases (Kiang et al., 1965).

The actual mechanisms) responsible for the decrement

during adaptation and for the postmasking decrement of the

probe-evoked responses is not known. The source of the

mechanism has been attributed by most investigators to the

hair cell-nerve fiber synapse (Rosenblith, 1954; Hawkins and

Kniazuk, 1950; Sorensen, 1959; Peake et al., 1962; Young and

Sachs, 1973; Spoor et al., 1976; Prijs, 1980). The hair cell

cochlear microphonics (CM) do not change with repetitive

stimuli (Sorensen, 1959; Peake et al., 1962; Young and Sachs,

1973; Charlet de Sauvage and Aran, 1976; Huang and Buchwald,

1980) nor do the summating potentials (Dallos, Schoeny, and

Cheatham, 1972; Eggermont, 1976a). The efferent olivocochleo

bundle (OCB) was considered a partial contributor by Liebrandt

(1965) who demonstrated that the AP decrement observed with

repetitive tone-bursts was abolished after injections of

procaine into the internal auditory canal. Liebrandt (1965)

suggested that efferent activity may play a part in several

portions of the central nervous system in causing a decrement

in evoked potentials following repetitive stimuli. Others,

however, do not consider the efferent pathways responsible for

the AP decrements since their effect on the AP requires much

higher rates of stimulation, stimulation does not cause any

reduction of spontaneous activity in auditory nerve fibers,

the AP decrements are still observed following complete

sectioning of the auditory nerve, and contralateral stimula-

tion does not produce any response changes (Sorensen, 1959;

Kupperman, 1972; Young and Sachs, 1973; Don, Allen and Starr,

1977; Huang and Buchwald, 1980). Middle ear muscle activity

has also been ruled out on the basis that the postmasking

effect is present at levels below middle ear muscle reflex

threshold, contraction of middle ear muscles does not affect

click-evoked APs, and the postmasking effect can be observed

after sectioning the muscles or following high doses of

anesthesia (Hawkins and Kniazuk, 1950; S'rensen, 1959; Liebrandt,

1965; Coats, 1971; Young and Sachs, 1973). The refractory

period of auditory nerve fibers may be important for simul-

taneous masking phenomena, but cannot explain the effects seen

in forward masking for the long ATs (Rosenblith, 1950; Hawkins

and Kniazuk, 1950; Peake et al., 1962; Harris and Dallos,

1979; Huang and Buchwald, 1980).

Recently, the Auditory Brainstem Response (ABR) has

become popular as a noninvasive far-field electrophysiological

recording procedure for assessing a short-latency system of

the auditory nerve and brainstem pathways (Jewett and Williston,

1971; Picton, Hillyard, Krausz, and Galambos, 1974; Buchwald

and Huang, 1975; Starr and Hamilton, 1976). The ABR can pro-

vide information on peripheral as well as more central response

capabilities. The ABR to temporally related stimuli has not

been as extensively investigated as has the AP. Repetition

rate and stimulus-train paradigms have been applied to

measures of the ABR and some of the brainstem components have

been shown to behave differently than the auditory nerve

response (Jewett and Williston, 1971; Picton et al., 1974;

Terkildsen, Osterhammel, and Huis in't Veld, 1975; Thornton

and Coleman, 1975; Pratt and Sohmer, 1976; Zllner, Karnahl,

and Stange, 1976; Hyde, Stephens, and Thornton, 1976; Martin,

1976; Weber and Fujikawa, 1977; Don et al., 1977; Kodera,

Yamada, Yamane, and Suzuki, 1978; Scott and Harkins, 1978;

Klein and Teas, 1978; Chiappa, Gladstone, and Young, 1979;

Kevanishvilli and Lagidze, 1979; Despland and Galambos, 1980;

Harkins and Lenhardt, 1980). These studies will be described

in more detail in a later section.

The different behavior of the various brainstem components

to temporally related stimuli is potentially quite interesting

and may indicate that peripheral auditory information is coded

in different ways in the brainstem. Response differences to

repetitive stimuli have also been noted for the various brain-

stem nuclei (Kiang, 1968; Winklegren, 1968; Watanabe and

Simada, 1969; Kitahata, Amakata, and Galambos, 1969; Kitzes

and Buchwald, 1969; Webster, 1971).

Knowledge of the postmasking recovery properties of the

ABR and their relation to postmasking recovery properties of

the AP may be useful as a method for objectively assessing

a dynamic property of the auditory system at more than one

level in a noninvasive manner. Provided that the normal

postmasking recovery properties are known, certain altera-

tions in clinical populations may be of diagnostic value as

has been suggested by Charlet de Sauvage and Aran (1976) and

Zllner, Karnahl, and Stange (1976) for cochlear pathologies,

by Yoshie and Ohashi (1971) for detecting acoustic neuromas,

by Robinson and Rudge (1977) for cases involving multiple

sclerosis, and by Salamy, McKean, Pettett, and Mendelson

(1978), Fujikawa and Weber (1977), Rowe (1978), and Despland

and Galambos (1980) for different developmental states. In

addition, postmasking recovery may be a method to better

assess persons who have poor speech discrimination beyond

that which can be due to their loss of pure-tone sensitivity

(phonemic regression), a characteristic of some geriatric

persons. Since continuous speech involves alternations of

intense vowels with relatively weak consonants, frequent

pauses and silent intervals, onsets and offsets with different

rates of amplitude changes, short-time spectral changes, and

different sound durations (Shoup and Pfeifer, 1976; Stevens,

1980), knowledge of temporal masking effects may have

important implications for better understanding these dynamic

characteristics of speech discrimination. Since most auditory

experience does not consist of simple sounds in a quiet

background, it is of interest to understand how one sound

affects the response to a simultaneous or subsequent sound.

At the single auditory nerve fiber level, Delgutte (1980)

suggests the importance of temporally related effects for

speech discrimination in the following:

Consider a region of rapid spectral change
in a speech signal (transition interval)
preceded by a steady segment having most of
its energy over a limited frequency region.
This segment would adapt most of the units
whose CF is in that region, so that these
units would be less responsive to the follow-
ing transition interval /-and_. this decreased
responsiveness would occur even if there
were a 50-100 msec silent interval between
the adapting segment and the transition
interval. (p. 847)

Temporal intervals up to 150 msec are typical between transi-

tions at the release of a stop consonant and the preceding

vowel (Delgutte, 1980). It is of interest also to understand

how these temporal effects are manifested in evoked potentials

and how they relate at different levels of the auditory system

in order to more fully understand basic auditory processing,

as well as, a potential clinical technique to help differentiate

certain abnormalities.

The purpose of this investigation is to provide some

normative descriptions of the postmasking recovery functions

for the ABR (wave V) in humans. A forward masking paradigm

is used to measure the click-evoked potentials for various ATs

following a noise-burst masker. AP responses from the ear

canal are also measured simultaneously with the ABR in some

of the subjects so that a direct comparison of the postmask-

ing recovery functions at two levels of the auditory system

can be made. Recent evidence suggests that under certain

stimulus conditions (especially repetition rate) and certain

pathological conditions, the brainstem responses do not simply

follow events occurring at the auditory periphery (Pratt and

Sohmer, 1976; Coats, Martin, and Kidder, 1979; Kevanishvilli

and Lagidze, 1979); therefore, one of the emphases of this

investigation is to compare the forward masking effects on

the ABR to the forward masking effects on the AP. Since

longer duration makers increase the postmasking effect (Coats,

1964a; Harris and Dallos, 1979), this type of forward masking

may have a more pronounced effect than repetition rate experi-

ments on the ABR and may prove in future applications to be

more useful in assessing certain clinical abnormalities at

different levels of the auditory system once the normal

characteristics are known. The postmasking recovery functions

are described in this investigation for different masker

durations, masker levels, and masker frequency bands (high-

pass masking).

General Description of the AP

The whole-nerve action potential (AP) represents a

population of synchronously responding auditory nerve fibers

to an abrupt acoustic stimulus and reflects both hair cell

function and activity in the auditory nerve (Dallos, 1973).

Recording of the AP can be achieved with electrodes on the

auditory nerve (Derbyshire and Davis, 1935; Goldstein and

Kiang, 1958; Kiang, Moxon, and Kahn, 1976; Huang and Buchwald,

1980), within the cochlea (Tasaki, Davis, and Eldredge, 1954;

Teas, Eldredge, and Davis, 1962), near the round window

(Rosenblith, 1954; Peake et al., 1962; Eggermont and Odenthal,

1974a,b,c; Dallos and Cheatham, 1976), or from the external

ear canal (Yoshie, Ohashi, and Suzuki, 1967; Cullen, Ellis,

Berlin, Lousteau, 1972; Elberling, 1974; Coats, 1974, 1976;

Berlin and Gondra, 1976; Kramer and Teas, 1979).

Auditory stimulation at high intensities produces an AP

response that consists of an initial negative deflection, labe

Nl, which occurs about 1.5-2.5 msec poststimulus onset and

followed by a smaller positive deflection, then negative again

forming N2 about 1 msec after N1 (Derbyshire and Davis, 1935;

Teas et al., 1962). In humans, the N2 is not as prominent

as in animals (Elberling, 1976a; Yoshie, 1976; Spoor et al.,

1976; Kramer and Teas, 1979). Most descriptions of the AP


response involve measurements of the latency and amplitude

of the N1 components as a function of stimulus intensity.

In general, as intensity increases the latency of N1 decreases

and the amplitude increases.

The most common stimulus used for evoking the AP is the

broadband click because of its rapid onset, which produces

a highly synchronous response from a population of nerve fibers

over a wide span of frequencies. Since estimates of the

basilar membrane traveling wave velocity indicate that the

velocity is greatest at the basal end and progressively de-

creases towards the apex (Teas et al., 1962; Zerlin, 1969;

Eggermont, 1975), the neural discharges from the nerve fibers

in the basal region occur sooner than those from apical regions

and are more synchronous because many discharges occur within

a brief time period. In response to broadband clicks, these

more synchronous basal fiber discharges predominate when

responses are recorded with distant electrodes, as is the AP

response (Teas et al., 1962; Schmidt and Spoor, 1974;

Eggermont, Spoor and Odenthal, 1976). More recently, it has

been demonstrated that tone-bursts and filtered clicks with

abrupt rise-times are good compromise stimuli for providing a

more frequency-specific evaluation along the cochlear parti-

tion at low intensities (Eggermont, 1976b; Zerlin and Naunton,

1976; Eggermont et al., 1976; Coats, 1976; Kramer and Teas,

1979; Coats, Martin, and Kidder, 1979). For higher in-

tensities, the basal extension of events on the basilar

membrane results in a more limited frequency specificity due

to an inclusion of the more synchronous high frequency re-

sponding populations (Teas et al., 1962; Elberling, 1974;

Eggermont et al., 1976; Kramer and Teas, 1979). A technique

to obtain responses from more frequency-specific regions

along the cochlear partition, even at high intensities, is to

add high-passed noise makers which can exclude the dominant

high frequency region and allow a probe response from more

apical regions (Teas et al., 1962; Elberling, 1974; Eggermont

et al., 1976; Zerlin and Naunton, 1976; Kramer and Teas, 1979).

The recording of the AP response "threshold" has been

demonstrated to correspond fairly well with conventional be-

havioral thresholds (Aran, Charlet de Sauvage, and Pelerin,

1971; Yoshie, 1973; Eggermont, 1975; Montandon, Shepard, Mass,

Peake, and Kiang, 1975; Elberling and Salomon, 1976) and

characteristic differences in waveform and the latency and

amplitude input-output functions have been found between

normals and pathological patients (Odenthal and Eggermont,

1974; Montandon et al., 1975; Yoshie, 1976; Elberling and

Salomon, 1976; Coats et al., 1979).

Forward Masking of the Whole-Nerve
Action Potential (AP)

Early evidence of forward masking of the AP can be found

in the work of Derbyshire and Davis (1935) who were the first

to demonstrate a decline in the AP voltage recorded from a

gross electrode on the auditory nerve in cats during tonal

stimulation, a phenomenon termed equilibrationn." From con-

tinuous photographic records, Derbyshire and Davis (1935)

measured the amplitudes of the APs over time for different

stimulating frequencies. Although the APs continued to decline

in voltage up to 10 minutes following the onset of the tone,

the largest and most rapid decline occurred during the first

2 seconds. The amount of reduction for the first AP follow-

ing the "onset" AP was found to be negligible for frequencies

below 400 Hz. Above a critical frequency of 1000 Hz the

voltage declined abruptly until about 4000 Hz where synchronous

APs were not detectable. Derbyshire and Davis (1935) sug-

gested that the decline of the first AP after the "onset" AP

was a reflection of the "functional" refractory period of

individual auditory nerve fibers (1.0 msec) and the continued

decline over the following 2 seconds was due to prolongation

of the relative refractory period. Galambos and Davis (1943)

demonstrated that the AP amplitude decline during continuous

stimulation was due, primarily, to a decline in discharge

rate of individual auditory nerve fibers. Maximum discharge

rate of single fibers averaged about 400/sec; however, dis-

charge rates up to 1000/sec were possible. A decline in dis-

charge rate (adaptation) to about 25% of the maximum rate

occurred within 1.0 sec. Coats (1964a) explained these

original findings of Derbyshire and Davis (1935) and

Galambos and Davis (1943) in terms of forward masking, where

each cycle of the stimulating tone had a forward masking

effect on subsequent cycles and the amount of decline de-

pended on the period of the tone as well as the duration of

stimulation. Derbyshire and Davis (1935) were also the first

to report on masking of the click-evoked APs by continuous

noise or tones. A demonstration of forward masking was

apparent in their results which showed that the largest mask-

ing effect occurred when the click response coincided with

the peak of a low frequency tonal masker (in phase)i however,

some masking of the click-evoked AP occurred for certain other

phase relations.

Rosenblith, Galambos, and Hirsh (1950) and Rosenblith

(1954), although primarily interested in investigation of the

AP following intense prolonged noise makers, observed that

the click-evoked AP from the round window in cats had a

sizable reduction in amplitude following even a short ex-

posure to moderately intense tones.

The first quantitative description of the postmasking

recovery of the AP was by Hawkins and Kniazuk (1950) who

recorded from the round window in cats click-evoked APs pre-

ceded by a 1.3 msec noise or tonal masker. These authors

found that full recovery of AP amplitude did not occur until

300 msec following a 50 dB SPL masker and 1.0 sec following

a 70 dB SPL masker. Masker duration did not affect recovery

for the 50 dB SPL masker, but delayed recovery was found as a

function of masker duration for masker intensities above 50

dB SPL. For tonal makers, Hawkins and Kniazuk (1950) re-

ported that 4000 Hz was the most effective masker of the

click-evoked APs and resulted in the greatest delayed re-

covery from masking. That the masking effects extended over

a period far in excess of the refractory period of auditory

nerve fibers was an important observation and Hawkins and

Kniazuk (1950) suggested that the long postmasking effects

involved some mechanism which transmits excitation from the

hair cell to the nerve fiber.

Using a paired-click paradigm in the cat, Rosenzweig

and Rosenblith (1953), Rosenblith (1954), and Sorensen (1959),

all demonstrated that the AP to the second click was reduced

in amplitude and the reduction was dependent on the interval

between the clicks as well as the intensity of the first

click. AP amplitude was still reduced for intervals far in

excess of the refractory period of the nerve fibers.

Sorensen (1959) investigated the forward masking effects

in guinea pigs using a low level (40 dB SPL) 1.0 sec noise

masker preceding a click presented at different levels. Post-

masking recovery functions were found to be dependent on

masker level as well as on probe level. Recovery time in-

creased as the signal-to-noise ratio decreased and ranged from

50 to 600 msec for complete recovery. Results with tonal

makers showed a much smaller effect than the white noise

makers and no significant differences were observed as a

function of masker frequency. Latency changes of N1 were not

reported. Sorensen (1959) suggested two possibilities to

explain the postmasking effects observed for the AP. Either

(1) the decrease in N1 amplitude is due to a decrease in the

number of active nerve fibers while each fiber maintains a

constant firing rate, or (2) the decrease in N1 amplitude is

due to a decrease in firing rate of all the active fibers

while the number of active fibers remains constant.

Peake et al. (1962) recorded AP responses to repetitive

stimuli (clicks and brief noise-bursts) in cats with a wire

electrode near the round window. Following the onset of

the stimulus train, the AP amplitude to successive stimuli

systematically decreased until some steady-state level was

reached. The APs to successive stimuli were observed and

it was found that the amount of decline increased as the

repetition rate increased up to 3000/sec. For rates above

3000/sec (ISI=.33 msec) no stimulus-locked activity could be

detected after the "onset" response. For rates below 5/sec

(ISI=200 msec) there was little change in amplitude for pro-

longed stimulation. The amount of amplitude decline measured

after the steady-state level had been reached was compared

for the different repetition rates and revealed constant

amplitude for rates up to 10/sec and then exponentially decayed

(except for a "bump" between 300 and 800/sec) until no neural

response was observed above a rate of 3000/sec. It was also

observed that for rates between 10 and 50/sec the entire AP

waveform decreased in amplitude without any change of N1

latency. Between rates of 50 and 300/sec there was a continued

decrease in amplitude as well as a change in latency of the

components later than N1. Above rates of 300/sec interpretation

was difficult because of response overlapping. Huang and

Buchwald (1980), recording from the auditory nerve and

cochlear nucleus using repetitive tone-bursts, found that

evoked potential decrements were parallel at the two loca-

tions. A complex interaction of repetition rate and stimulus

duration was reported and was determined to be due to the

intertone interval. For intertone intervals greater than

100 msec, evoked potential decrements were not observed

regardless of stimulus duration and repetition rate.

In a series of articles, Coats (1964a,b, 1967, 1971)

described the time courses of AP recovery following tonal

and white noise makers in cats for different masker durations,

masker intensities, and probe-click intensities. The shortest

AT measured was 50 msec. Increasing the masker duration from

.3 to 3.3 sec resulted in a decrease in the AP recovery rate

(longer recovery with longer duration). Changing the duration

of the masker was found not to have any effect on the masked

AP amplitude under simultaneous masking conditions; therefore,

the "masking duration effect" acted only to slow the rate of

recovery following the masker. Increasing masker intensity

was shown to increase the amount of AP depression (amplitude

reduction), but the recovery functions had parallel time

courses until AP amplitudes were above 80% of control. In

addition, Coats reported that the maskerr duration effect"

was minimal for masker intensities below 50 dB SPL and in-

creased as a function of masker level above 50 dB SPL. In

contrast to the maskerr duration effect," the maskerr

intensity effect" was related to decreasing the AP amplitude,

but had no effect on the rate of recovery. Varying click

intensity was similar to the effect of changing masker in-

tensity on the amplitude of the AP. The latency of N1 was

found to change only slightly under the forward masking


Kupperman (1971) found a close comparison between human

and guinea pig AP depression as a function of click repetition

rate except that the amount of amplitude reduction was less

pronounced in humans.

Yoshie (1968) investigated changes in N1 amplitude as a

function of repetition rate in humans using a needle electrode

inserted into the wall of the external ear canal. The clicks

were presented at 70 dB sensation level for rates of 1-200/sec.

Averaged steady-state amplitude remained unchanged up to

rates of 10/sec. A monotonic decline in amplitude was found

for rates above 10/sec, but the latency of N1 and the wave-

form configuration remained unchanged. Using a paired-click

paradigm, Yoshie (1968) measured AP amplitude as a function

of ISI and found that full recovery occurred after 100 msec,

which corresponded to the critical value found in the rate

function (10/sec). Yoshie and Ohashi (1971) compared normal

AP responses to click-trains to those obtained from a

patient having bilateral acoustic neuromas, moderate

sensorineural hearing loss, abnormal tone decay, and poor

speech discrimination. For the normal subjects, the ampli-

tude decrease across time was very rapid over the first

10-20 msec of the click-train followed by a more gradual

decrease. The amount of amplitude decline increased as ISI

decreased. In contrast, the patient showed a slower initial

amplitude decline and continued to decline over a longer time

period, up to 100 msec, i.e., no clear distinction could be

made between the early rapid decline and the later slower

decline as was seen in the normals. The locus of the abnormal

AP responses to the repetitive stimuli could not be estab-

lished, however, the authors suggest that the abnormal AP

decrement could be due to a mechanism of sensory cell-nerve

fiber synapse reflecting sensory cell damage as well as

properties of the nerve fibers.

Eggermont and his coworkers (Spoor, 1965; Eggermont and

Spoor, 1973a,b; Eggermont and Odenthal, 1974a; Spoor et al.,

1976) have provided an extensive description of forward mask-

ing effects on the AP using noise makers, stimulus-trains,

and repetition rates. They used 4000 and 6000 Hz tone-bursts

as the probe stimuli. The recovery functions under all three

conditions followed an exponential time course. Changes

were observed both in N1 latency and amplitude as a function

of repetition rate and as a function of AT; however, the

recovery took longer in the noise masking case due to the

duration effect (Coats, 1964a). Following the 500 msec

noise masker, the N1 amplitude reached full recovery after

250 msec in guinea pigs while the latency was fully recovered

by 64 msec. Masker levels below 60 dB SPL produced lesser

amounts of AP decrement at masker offset, but the slope of

the recovery functions became less steep as masker intensity

decreased. For masker levels above 70 dB SPL, the time for

full recovery increased to values beyond 250 msec. In addi-

tion, it was shown that N1 recovery was about 4 times longer

(up to 1000 msec) in man than in guinea pigs using the same

procedure. This was suggested to be due to different proper-

ties at the hair cell-nerve fiber synapse between the two

species. These authors also suggested that the forward mask-

ing effects of the AP are caused by adaptation properties

of each active individual nerve fiber resulting in a reduction

in discharge rate. The adaptation, in turn, allows spon-

taneous desynchronizing factors (synapse or membrane noise)

to contribute to the latency distribution function (Goldstein

and Kiang, 1958) and produce an increase in latency (because

of a broadening) and a decrease in N1 amplitude. The number

of potentially active nerve fibers remains the same and is

determined by the stimulus intensity and frequency. These

authors further suggest that the N1 recovery curve can only

be influenced by the refractory properties of the auditory

nerve fibers up to about 20 msec. For longer times, the

postmasking effects are primarily due to adaptation mechanisms.

Charlet de Sauvage and Aran (1976) investigated AP

responses to click-trains in normal adults and in patients

with various sensorineural hearing abnormalities. They used

a train of 5 clicks with an ISI of 8.5 msec and 100 msec

between click-trains. Only amplitudes were measured since,

in contrast to Eggermont and coworkers, they did not observe

any N1 latency shifts for the responses. (They attributed

this difference perhaps to the fact that clicks were used

instead of tone-bursts which were used by Eggermont and

coworkers.) Charlet de Sauvage and Aran (1976) showed that

AP decrements were most pronounced when the click-trains were

at 60 dB Hearing Level (HL), corresponding to the "knee" of

the input-output function. These authors suggested that if

the response amplitude of the 4th click is greater than 76%

or less than 36% of the response to the 1st click this should

be considered abnormal; however, they found that whenever

the response decrement was abnormal it was to the side of

being greater than 76% of the first click, i.e., less re-

sponse decrement. Response decrement of the click-trains

was found to be significantly different from normal only for

dissociated and recruiting ears. These authors suggested

that in those pathological ears which showed abnormal response

decrements there were, primarily, outer hair cell losses.

Since normals showed a much larger response decrement and the

decrement was maximal at 60 dB (the hypothetical point of

changeover from outer hair cell to inner hair cell response)

the pathological ears were apparently missing this mechanism.

In summary, it appears that a (simple) sound can clearly

affect the normal AP response to a following sound in a fairly

dramatic manner, i.e., a significant reduction in response

amplitude occurs for temporal intervals at least up to 100

msec. The effect on N1 latency is not entirely agreed upon.

The extended masking effects occur for temporal intervals

far in excess of refractory mechanisms. Recovery functions

at the underlying single fiber level closely parallel those

of N1 and provide evidence that a reduction in firing rate

of each individual fiber following a normal adaptation process

is a basic mechanism of forward masking at this level of the

auditory system. Since the AP is a population response, one

must consider that the sum total of the contributing single

fibers may be manifested in the AP in a complex fashion.

In addition, one cannot rule out the possibility that a re-

duction in number of nerve fibers plays a role in the

observed effects on the AP under forward masking conditions.

The source of the forward masking effects is not known; how-

ever, it is most likely related to a transduction mechanism

at the hair cell-nerve fiber synapse, and this basic normal

mechanism may be deficient in certain pathologic cases.

General Description of the ABR

The auditory brainstem response (ABR) was first described

by Sohmer and Feinmesser (1967) who observed a series of

potentials following N1 when recording from the external ear

canal and attributed them to brainstem sources. Soon after,

Jewett (1969, 1970),recording in cats, observed four positive

waves occurring between 2-7 msec following a click stimulus.

By simultaneously recording at various brainstem locations,

wave 1 was found to be synchronous with the auditory nerve

and wave 2 to activity of the cochlear nucleus. Regions

corresponding to waves 2 through 5 were found to have fast

and slow components, and since the slow components have longer

latencies, they could add to the activity from higher auditory

centers. Nevertheless, wave 3 was attributed primarily to

the superior olivary complex, wave 4 between the lateral

lemniscus and the inferior colliculus, and wave 5 to the in-

ferior colliculus. Similar conclusions were found by Buchwald

and Huang (1975), Huang and Buchwald (1977), and Achor and

Starr (1980) using ablation and depth recording techniques

in cats. Starr and Hamilton (1976) recorded ABRs in human

patients with confirmed brainstem pathologies and their

results indicated ABR sources which were in agreement with

those of previous investigators.

The most prominent and stable wave of the ABR is wave V,

which is the current wave of interest in most auditory in-

vestigations of the ABR. In response to a click at high

intensities, wave V shows a characteristic latency between

5.2 and 6.2 msec and can be reliably recorded from the

same subject over time (Jewett and Williston, 1971; Picton

et al., 1974; Starr and Achor, 1975). Schulman-Galambos and

Galambos (1975), Hecox and Galambos (1974), Salamy et al.

(1978), and Despland and Galambos (1980) demonstrated that

wave V was present by 30 weeks gestational age and that the

latency of wave V decreased with infant maturation until the

age of 12-18 months, at which time wave V latency was the

same as adult values. The ABR has also been shown to be

independent of level of consciousness or stage of sleep

(Amadeo and Shagass, 1973; Picton and Hillyard, 1974; Uziel

and Benezech, 1978) or levels of anesthesia (Starr, Hamilton,

and Achor, 1974; Gerull, Biesen, Mrowinski and Rudulph, 1974;

Bobbin, May and Lemoine, 1979).

Amplitude measures of the ABR have not been reported as

often as have the latencies of the waves. Amplitudes of the

ABR are very small (nanovolt range) and most investigators

have shown a large variability across subjects and within the

same subjects (Picton et al., 1974; Starr and Achor, 1975;

Rowe, 1978; Chiappa et al., 1979).

Investigation of the ABR centers, primarily, on the

relation of the latencies (especially wave V) to different

stimulus variables, e.g., the latency of wave V has been shown

to systematically decrease as the stimulus intensity is in-

creased. Stimulus frequency (Davis, 1976; Klein and Teas,

1978; Coats et al., 1979) and repetition rate (Jewett and

Williston, 1971; Pratt and Sohmer, 1976; Harkins et al., 1979)

also affect the ABR. The extent that the various ABR com-

ponents reflect activity occurring at the periphery is an

interesting question. This is extremely important if one

expects to use, say wave V, to approximate behavioral audio-

grams. The suggestion that wave V is closely linked to the

synchronous volley of nerve impulses occurring at the

periphery, except displaced in time has been made by some in-

vestigators (Davis, 1976; Elberling, 1976a;Klein and Teas,

1978; Kramer and Teas, 1979). However, some evidence exists

which indicates that wave V behaves differently than N1

under certain conditions, e.g., as a function of stimulus fre-

quency, repetition rate, or types of hearing loss (Thornton

and Coleman, 1975; Pratt and Sohmer, 1976; Kevanishvilli and

Lagidze, 1979; Coats et al., 1979; Kramer and Teas, 1979).

Certainly, peripheral activity serves as input to the brainstem

and therefore different contributing response areas along

the cochlear partition will produce different inputs. The

frequency specificity of wave V should be influenced by

the changing spatial distribution of N1. To more closely

partition out various spatial (and frequency) distributions,

high-pass masking has also been applied to the ABR and results

have indicated a close correspondence of wave V and N1

latency across frequency region (Don and Eggermont, 1978;

Kramer and Teas, 1979). However, the amplitude of wave V

remained fairly stable across frequency regions while N1

amplitude decreased for lower frequency regions (Don and

Eggermont, 1978; Parker and Thornton, 1978).

The relation between stimulus variables and latency is

precise enough to make the ABR useful for accurate

estimations of auditory thresholds (Davis, 1976; Galambos

and Hecox, 1977; Pratt and Sohmer, 1976), for evaluation of

auditory function in infants (Despland and Galambos, 1980),

and in the evaluation of neurological disorders (Stockard

and Rossiter, 1977; Starr and Achor, 1975; Stockard, Rossiter,

Wiederholt, and Kobayashi, 1976; Robinson and Rudge, 1977;

Stockard, Stockard, and Sharbrough, 1977; Shanon, Gold,

Himmelfarb, and Carasso, 1979; House and Brackman, 1979).

Forward Masking of the Auditory
Brainstem Response (ABR)

Interest in the effect of temporally related stimuli on

the ABR has been restricted to repetition rate or stimulus-

train paradigms. Generally, the concern has been for identify-

ing a repetition rate most applicable for routine clinical

testing, i.e., establishing a repetition rate that provides

a clear and stable response within the shortest time period.

Only recently have the effects of repetition rate per se been

quantitatively investigated in terms of assessing a dynamic

property of the brainstem pathways, with the implication

that certain disease states may cause a failure in adequate

conduction of impulses when driven at high repetition rates.

Jewett and Williston (1971) first demonstrated the

effect of click repetition rate on the ABR in normal hearing

adults. They reported that wave V showed no amplitude or

latency shifts as repetition rate increased from 2.5 to

50/sec, while earlier peaks were severely reduced in ampli-

tude, but showed no latency shifts. Wave V amplitude was

even shown to increase slightly at the higher repetition

rates. It was suggested that wave V could be a useful candi-

date for clinical measures since faster repetition rates could

be employed without affecting the amplitude of the response.

Picton et al. (1974) confirmed the findings of Jewett and

Williston (1971) for click repetition rates up to 60/sec.

Terkildsen, Osterhammel, and Huis in't Veld (1975) re-

corded ABRs in normal adults using tone-bursts (1,2,5 kHz)

at repetition rates of 5-40/sec. They found equal increases

in latency for waves I and V, but the amplitude of wave V

remained stable for the higher repetition rates while wave

I's amplitude was significantly reduced even at a repetition

rate of 10/sec. The effects of repetition rate were found

to be independent of tone-burst frequency. These authors

interpreted their findings as evidence for longer lasting

unit responses at the source of wave V, making wave V less

susceptible to desynchronizing factors.

A click-train paradigm was used by Thornton and Coleman

(1975) where ABRs were recorded to each click in the train

for ISIs of 15, 24, and 32.5 msec. These authors found that

the amplitude and latency changes of all the waves were

nearly complete by the 4th or 5th stimulus and the amount of

amplitude reduction increased with decreasing ISI and de-

creasing stimulus intensity. Wave V was found to show the

least amount of amplitude reduction compared to the earlier

waves. In addition, Thornton and Coleman (1975) reported that

the latency increases were smaller for the smaller amplitude

reductions, i.e., wave V showed the smallest latency increase

and wave I showed the largest latency increase. Thornton

and Coleman (1975) suggested that the decrement in wave I

amplitude is due to a reduction in firing rate of all active

nerve fibers while the number of nerve fibers remains con-

stant. In the brainstem, the reduced firing rate of incoming

fibers resultsin fewer responding cells because some cells

may not reach an adequate threshold level. This should re-

sult in amplitude decrements at each brainstem level parallel

with (or greater than) wave I. However, the amplitude

decrements become less at more rostral brainstem levels be-

cause of an increasing number of fibers and complex inter-

connections at each successive stage which results in a smaller

number of adapted outgoing fibers than the number of adapted

fibers at the input.

Pratt and Sohmer (1976) found, in humans, that as

click rate increased from 5 to 80/sec wave I amplitude

was significantly reduced, but showed no latency change,

and wave V latency increased, but showed only small ampli-

tude changes. The latency change of each wave was found to

be greater than the wave preceding it (accumulative). These

authors suggested that each wave is influenced by preceding

waves and "at a particular nucleus a smaller number of in-

coming fibers would be active and their synapses less effec-

tive, so that it would take longer for the excitatory

synaptic potential to reach the threshold of firing of

outgoing fibers" (p. 90). The amplitudes of later waves

remain stable because of increasing number of neurons

at higher levels which receive input from a smaller number

of lower-order fibers (divergence) as well as each higher-

order neuron receiving many endings from lower-order fibers


Don, Allen, and Starr (1977) studied the latency shift

of wave V in normal adults using click repetition rates of

10, 30, 50, and 100/sec at levels up to 60 dB HL and found

significant latency increases (up to 0.9 msec) for repetition

rates from 10 to 100/sec. This latency shift was found to

be equivalent to a 15-20 dB reduction in the stimulus level

and these authors felt that repetition rate should be con-

sidered carefully in interpreting response latencies. These

authors described the latency shifts as a linear function of

repetition rate, which they point out as being different

from the exponential shift of the AP latency found by Egger-

mont and coworkers. The different behavior of wave V compared

to N1 of the AP was interpreted by Don et al. (1977) to

possibly suggest different peripheral and central responsive-

ness to repetitive stimuli due to adaptation mechanisms along

the monaural pathway. However, Don et al. (1977) felt that

the locus of the wave V effects was at the auditory periphery.

Unfortunately, the resolution of waves I-IV was not clear

enough in the study of Don et al. (1977) to make any direct

comparisons of peripheral and central effects. The latency

increases with repetition rate were also found to be inde-

pendent of stimulus intensity. In another part of the same

study, Don et al. (1977) used click-trains with ISIs of 10

msec and demonstrated that the latency shift of wave V was

complete by the 5th stimulus. No latency shift of wave V

was observed in the click-train experiment when the ISI was

100 msec. A binaural central mechanism was ruled out in a

third experiment where they found no latency change to the 1st

and 20th click presented to the ipsilateral ear while the

2nd through 19th click were presented to the contralateral

ear (ISI=10 msec).

As the above studies show, there does not appear to be

complete agreement on the effects of increasing repetition

rate for the various components of the ABR. The effects of

click repetition rate on the ABR in normal adults have been

extensively studied by several other investigators for

repetition rates ranging from 2/sec to 100/sec (Pratt and

Sohmer, 1976; Zollner, Karnahl, and Stange, 1976; Hyde,

Stephens, and Thornton, 1976; Klein and Teas, 1978; Harkins,

McEvoy, and Scott, 1979; Martin, 1976; Weber and Fujikawa,

1977; Scott and Harkins, 1978; Chiappa, Gladstone, and Young,

1979). In general, the majority of these studies have shown

that the greatest effect of increasing click repetition rate

has been a large reduction of wave I amplitude and a sub-

stantial (up to about 0.9 msec) latency increase of wave V.

In contrast to the study by Don et al. (1977), most studies

describe the latency change of wave V as being a logrithmic

function of repetition rate. In contrast to Thornton and

Coleman's (1975) click-train study, most investigators report

that wave I latency does not increase or increases only

slightly at the higher repetition rate. Most investigators

agree that there is a cumulative latency increase for

successive peaks after wave I as a function of repetition

rate, suggesting that the latencies of successive waves are

influenced by the preceding wave. In contrast to Don et al.

(1977) and Thornton and Coleman (1975), the studies by

Zollner et al. (1976) and by Weber and Fujikawa (1977) have

shown that the effects of repetition rate are dependent on

stimulus intensity, i.e., greater effect at higher inten-

sities. The relatively large wave V at higher repetition

rates has enhanced the popularity of using wave V for ob-

jective clinical audiometry. However, as suggested by Zollner

et al. (1976), the examination of the waves earlier than

wave V, especially wave I, as a function of repetition rate

may be a more useful indicator of inner ear damage.

Bauer, Elmasian, and Galambos (1975) compared the

psychophysical loudness judgements with human ABRs for the.

second of two 0.2 msec 8 kHz tone-bursts separated by intervals

ranging from 100-400 msec. The psychophysical results showed

the expected "loudness enhancement" for some of the conditions

above 15 dB, but wave V latencies and amplitudes were not

affected under any of the conditions. The total response

power of the ABR was reduced which indicated a decrease in

the amplitude of the waves earlier than wave V since wave V

voltage was not altered (and loudness was enhanced).

These results suggest that the ABR does not correspond to

perceived loudness under certain conditions.

Kodera et al. (1978) compared the effects of repetition

rate on the ABR and the middle latency components using 500

and 1000 Hz tone-pips in order to determine which response

would be more suitable in routine clinical testing. It was

observed that as ISI decreased from 104 to 32 msec the

middle conponent's amplitude decreased significantly, while

wave V was not appreciably affected except for a slight

increase in amplitude at ISI=32 msec due to superpositioning

of responses from preceding stimuli. Latency increases were

found only for wave V as ISI decreased.

Huang and Buchwald (1980), on the other hand, showed

that in cats the amplitude of the later ABR waves (P3, P4)

decreased by the same amount as the earlier waves (PI, P2)

when repetition rate was increased. The amplitude de-

crease was gradual up to a repetition rate of 20/sec

then decreased sharply for rates up to 200/sec. All waves

showed a slight (0.1 msec) latency increase as repetition rate

was increased from 50 to 200/sec. Increasing the stimulus

duration beyond 10 msec caused an interaction with repetition

rate, i.e., using 100 msec tone-bursts at a rate of 10/sec

produced a larger amplitude reduction than seen for the

short duration stimuli at 10/sec. Duration-rate inter-

actions have also been shown to affect the wave V amplitude

in humans (Hecox et al., 1976).

In monkeys, Allen and Starr (1978) found no latency

shift of wave I, a significant (p < .05) latency shift of .1

msec for wave II, and 0.4 msec in wave IV (comparable to

wave V in humans) between click repetition rates of 10 and

100/sec. Significant interaction was found between rate and

stimulus intensity, i.e., at higher intensities wave IV

showed larger latency increases with increasing repetition


Kevanishvilli and Lagidze (1979) used a response con-

ditioning paradigm to study the temporal effects on the ABR

in humans. The conditioning click and the test click were

separated by ATs of 5 or 10 msec. By subtracting the re-

sponse to the conditioning click from the response to the

conditioning click-plus-test click, the contamination by

middle latency components and postauricular potentials was

eliminated. Their results showed, for a AT of 10 msec, that

wave I was significantly reduced in amplitude while wave V

was not affected. The latency changes of wave I and V were

the same. For AT=5 msec, wave V latency increased much more

than wave I latency. These authors attributed their find-

ings to peripheral desynchronizing factors which have

different effects on various waves depending on the durations

of neural contributions to each wave, i.e., wave V shows less

amplitude decrease because of its relatively longer response

component which makes it less susceptable to desynchroniza-


The effects of repetition rate on the ABR as a function

of age were first reported by Jewett and Romano (1972) for cats

and rats during the first 2-3 months of life. Young kittens

were found to be different from adult cats and young rat

pups in that the kittens did not show any responses to 100/sec

clicks until several days after ABRs were detectable at slower

repetition rates. Soon after 100/sec responses could be ob-

tained in the kittens (about 16 days old) subsequent stimula-

tion with clicks at 10/sec resulted in a lack of response

for several minutes, i.e., there was a "fatigability" factor

in the young kittens. For the rat pups and older cats,

stimulation at 100/sec produced decreased amplitude and in-

creased latency, which the authors suggested was due to some

manifestation of a "recovery cycle."

In humans, the effects of repetition rate on the ABR as

a function of age have not been firmly established. Fujikawa

and Weber (1977) compared wave V latency shifts as a function

of repetition rate in populations of 7-8 week old infants,

young adults, and geriatrics. All groups showed a latency

shift of wave V as repetition rate was changed from 33-67/sec,

and the amount of latency shift for the young adults and in-

fants was the same except for 67/sec where the infants showed

a greater latency shift. In contrast, the geriatric group

showed considerably longer wave V latencies than the adults

and infants at all the rates tested. The different behavior

of the geriatrics was attributed to a possible reduction in

the number of nerve fibers and/or cell count, whereas the

different behavior of the infants (at the highest rate) may

reflect a myelination factor. Salamy et al. (1978), on

the other hand, found that for infants from birth to 6 months

of age, the effects of increasing click repetition rate

from 10-80/sec (decreased amplitude of earlier waves and

increased latency of wave V) was not significantly different

from adults except for the newborns (24-27 hours). This

finding suggested that the mechanisms responsible for rate

effects are mature very shortly after birth. Despland and

Galambos (1980) found that responses to clicks up to 80/sec

could be measured in premature infants above 30 weeks

gestational age and that the latency shifts of wave V as a

function of repetition rate decreased with age from 270 usec

for each 10/sec increase in rate at the earliest age, down

to about 35 usec for each 10/sec increase in rate found in

adults. Wave I was also found to increase in latency as a

function of repetition rate; however, the amount of latency

shift was about one-half that seen for wave V. The rate

function of wave I across age paralleled wave V. These

authors state that one-half of the rate effects occur at

the periphery and one-half in the brainstem.

The effects of repetition rate on the ABR in clinical

patients have also not been extensively studied. Stressing

the system by increasing repetition rates (or other tem-

porally related stimuli) could be useful for differentiating

certain central brainstem pathologies. Stockard and Rossiter

(1977) and Stockard, Stockard, and Sharbrough (1977) found

that in patients with multiple sclerosis, a demyelinating

disease of the central nervous system, the ABRs to clicks at

10/sec were abnormal in all of them and showed increased

latency and/or reduced amplitude of various waves. Increasing

the repetition rate to 25-30/sec enhanced the sensitivity of

the test, i.e., greater response abnormality was found in

the patients with multiple sclerosis than could be accounted

for by the normal rate dependent changes. Since all the

patients had confirmed multiple sclerosis and all showed

abnormal responses at the 10/sec rate, there did not appear

to be any additional information obtained from the higher

repetition rate. Robinson and Rudge (1977), however, did

find that recording the ABR at high repetition rates was

useful in detecting and confirming patients with multiple

sclerosis. A unique procedure was used by Robinson and

Rudge (1977) for stressing the auditory system. The ABRs

were recorded to the first of a pair of clicks separated by

5 msec. In normals, when the pairs of clicks were presented

at a rate of 2.5/sec the latencies and amplitudes of all the

waves were the same as to a single click. When the pairs

of clicks were presented at a rate of 20/sec the response to

the first click showed significant increases in latencies of

waves I, III, and V, but wave I latency did not change

(changing repetition rate of single clicks from 2.5 to 20/sec

showed an increase in wave III latency only). Using this

paired-click paradigm presented at 20/sec in patients with

multiple sclerosis, Robinson and Rudge (1977) found that 55%

of those patients who had abnormal latencies to single clicks

and 57% of those patients who had abnormal amplitudes to

single clicks (and normal latency) showed an abnormal increase

in wave V latency to the paired-clicks. More importantly,

three patients who had normal ABRs to single clicks showed

either an abnormal increase in wave V latency or no response

at all to the first click of the pair. These authors suggest

that stressing the auditory system by combining paired-clicks

with a high rate of presentation (when either alone produced

no change) can reveal changes in the ABR which can be a

useful indicator of patients with unconfirmed or suspected

multiple sclerosis, even when normal ABRs or normal latencies

with reduced amplitudes are found to single clicks (20/sec).

In summary, the effects of preceding sounds on the ABR

response to another sound have been shown not to simply

parallel those changes that occur for the AP response. The

most obvious difference has been a stability of wave V ampli-

tude concomitant with a large increase in wave V latency,

whereas N1 has shown large amplitude reductions with only

slight latency changes. These differential manifestations

to repetitive stimuli at two levels of the auditory system

preclude the simple assumption that the perception of one

sound preceded by another sound is entirely due to peripheral

mechanisms which are "relayed" unchanged to higher centers.

Because ABRs do not always produce clear and stable waves

earlier than wave V, comparisons between waves are often

difficult to make. By simultaneously recording auditory


nerve activity from the ear canal, more direct comparisons

can be made. Further investigation of the effects of

temporally related sounds at different levels of the auditory

system appears warranted. Once the normal characteristics

are known more fully, application to clinical populations

may have more success in understanding normal auditory pro-



Click-evoked ABRs and APs were recorded using a forward

masking paradigm for delay times (AT) of 0 to 100 msec.

Responses obtained, as a function of AT, following a 60 msec

wide-band masker (.02-10 kHz) set at a level sufficient to

"fully" mask wave V of the ABR (AT=0 msec) are called the

"standard" series. In addition to the "standard" series,

responses were obtained (as a function of AT) for variations

of masker duration, masker level, and masker low frequency



Fifteen normal hearing (ANSI, 1969) adults, 23-33 years

of age, served as volunteer subjects. All 15 subjects

received the "standard" series. For each variation in

masker, i.e., duration, level, or low frequency cutoff, 7

subjects were used. Six subjects were willing to return for

a second recording session, providing data for two different

variations of masker.



The probe stimuli were 0.1 msec broadband clicks

generated by a shift-register pulse output with the duration

of the clicks controlled by a precision one-shot (logic

level). Intensity calibration for the probe clicks was based

on standard reference levels which have been in use at this

laboratory for the past 5 years. These levels were obtained

by determining the behavioral threshold to the clicks presented

at a repetition rate of 5/sec for a group of normal hearing

listeners. This was considered 0 dB HL (Hearing Level) to

which other intensity levelsware referenced. Calibration

levels were determined by measuring the voltage (and converting

to dB SPL) of the clicks at 60 dB HL presented at a repetition

rate of 250/sec on a true RMS voltmeter (Ballantine, 320).

Corresponding peak-to-peak voltage was measured on an oscillo-

scope (Tektronix, RM561A). Calibration levels for the 60 dB

HL broadband clicks were 72 dB SPL and .07 volts peak-to-peak.

An attenuator (Hewlett-Packard, 350D) provided control of the

click level as measured on the true RMS voltmeter. For this

investigation, the probe click intensity level was always

set at 50 dB HL.

The masker stimuli were obtained from a random noise

generator (Grason-Stadler, 455C). The duration of the noise,

onset-to-offset, was determined by another precision one-

shot and shaped by an electronic switch to produce nonlinear

rise-fall times of 10 msec. An attenuator (Hewlett-Packard,

350D) independently controlled the intensity of the noise

masker. The bandwidth of the noise was produced by feeding

the white-noise output of the noise generator into two

variable bandpass filter networks (Rockland, 816) wired in

series. Each bandpass filter had a rolloff of 48 dB/octave

with a resultant rolloff of 96 dB/octave. The high frequency

cutoff was always set at 10 kHz. The low frequency setting

of the noise could be changed to produce the desired makers,

e.g., .02-10 (referred to a wide-band), 2-10, 4-10, and 6-10

kHz. The frequency characteristics of the earphone (TDH-39)

produced additional filtering of the signals. The acoustic

outputs were measured in a standard 6-cc coupler with a

condenser microphone (Bruel and Kjaer, 4132) connected to a

sound level meter (Bruel and Kjaer, 2203). Spectral analysis

was performed by a software Fast Fourier Transform (FFT)

program. The noise spectra and earphone characteristics

are shown in Fig. 1.

A block diagram of the stimulus generation and masker-

probe timing relation are given in Fig. 2. To produce the

0 0
ft\ \0

NOIlVnN3iIV gp




-.T >-



*H 4-1

0 -
u 4-

On o
c 0


0 40


--I 4-

-U 4J

4 0


4 .
4J 4-i

4-4 Q)


Cn 0
H (U2
e n



m C

(H E

1 0)
/2 0
(U 4-

cn <






c "




V/ L)
- t

a i
0 L
*- -n vi
OE -

z C
-0 0

- C 4-'
'0 c

c a- L
n 0 1-1 in

C G.

0 0

-C 0J C-



E- E

01 -- C

E m1 o
0)- O
*- U 0
- o0 01
(U4o i

_ __ ___~_

forward masking paradigm, the click delay relative to the

masker was varied by another precision one-shot. The

delay time from the masker offset to the click onset was

defined as AT. The values of AT used were 0, 6.2, 12.5,

25, 50, 100 msec. These were determined by observation on

the oscilloscope display and setting of the one-shot. A

synch pulse (97 usec) to the computer was initiated at the

probe click onset. The masker-probe pairs were presented at

a repetition rate of 4/sec which resulted in an intermasker

interval (IMI), i.e., time between offset of one masker

and onset of subsequent masker, of 190 msec. For the series

where masker duration was varied, the repetition rate remained

constant (4/sec) resulting in different IMIs for each duration.

For the maximum AT (100 msec), the following masker occurred

90 msec later. This reduced any possible backward masking

effects. The probe and makers were fed sequentially to a

mixer, an amplifier (James B. Lansing, SE 400S), a final

attenuator (Hewlett-Packard, 350D), and then monaurally

presented through the earphone. The equipment allowed for

the presentation of the signal or noise separately or the

signal presented with (after) the noise. The signal and noise

levels could be varied independently or simultaneously.


Three silver/silver chloride scalp electrodes were used

for the ABR recordings. The electrical activity was recorded

between forehead and ipsilateral mastoid with the contra-

lateral mastoid serving as ground. The electrode sites were

abraded with electrode paste (Grass, EC-2) then taped to the

forehead and mastoids. Electrode resistance was checked with

a VOM meter and the electrodes were reattached if the resis-

tance was greater than 5 k Ohms.

The ear canal recording required a specially made elec-

trode. This was made of 6 inches of teflon coated silver

wire (diameter=.007 inches). The teflon was scraped off the

end and the stripped end of the wire then wrapped around the

end of a narrow (1/8th inch) piece of polyurethane surgical

rubber rubing (about 3/4 inch long). The tubing was then

bent in a "V" before inserting into the ear canal. After

the electrode was positioned inferiorly in the ear canal,

with the aid of an otoscope, earlight, and forceps, the tubing

was released so that it opened slightly to hold itself in

place against the walls of the ear canal. Good responses

were obtained only when the wire wrapped end of the ear canal

electrode was in contact with the canal wall. If poor re-

cordings were obtained the ear canal electrode was repositioned,

but if poor responses were still obtained after one or two

attempts at repositioning no further attempts were made and

ear canal recordings were not obtained. The reference

electrode for the ear canal recordings was the ipsilateral

mastoid and the ground was the contralateral mastoid.


The subject laid on a bed in a sound treated room (IAC).

Two recording channels were used for independent monitoring

of the ABR and the AP. Each electrode configuration was

connected to a separate high-impedance cathode follower

(Grass, HIP 511E) and then to separate a.c. coupler high-

gain preamplifiers (Grass, P511). The gain of the amplifier

was set at 100,000 and bandpassed at 0.03-3 kHz.

The electrical activity from the ABR and ear canal

electrodes was recorded on separate channels of an FM tape

recorder (Ampex, FR 100A) bandpassed at 0-2500 Hz at 7 1/2

i.p.s. and kept for later off-line processing. Additional

response filtering (Krohn-Hite, 330M) was done ahead of the

computer with a bandpass of 70-1600 Hz. The subject's

electrical brain activity was also monitored on the oscillo-

scope. A general purpose computer (Digital Equipment,

PDP-8e) was used for the response averaging. Each response

represented the averaged stimulus-locked activity for 1200-1500

samples. A sweep duration of 10.0 msec (39 usec/point, 256

points) was used for the ear canal response. For the ABR,

a sweep duration of 20.2 msec (79 usec/point, 256 points)

was used. A toggle switch allowed processing of either the

ABR or the AP activity. However, after determining whether

or not the ear canal electrode was working properly, only

the ABR was monitored on-line and the averaged ABRs were

stored directly on floppy disks. The AP responses from the

ear canal were played back later off the FM tape and then

stored on the floppy disks. A feature of the averaging pro-

gram was variable artifact rejection, which was set by two

independent cursor lines. In these experiments, electrical

activity greater than 7 pvolts was rejected from the average

by the computer. This criterion excluded unwanted muscle or

movement artifacts from the averaged response.

In off-line processing, latencies and peak-to-peak ampli-

tudes were obtained by means of a cursor which could be set

at any location in the waveform and read from the monitor


Each session began with the "standard" series which con-

sisted of a 60 msec wide-band (0.2-10 kHz) noise masker pre-

ceding the broadband probe click (50 dB HL). The level of

the wide-band masker was set (and measured) so that, with a

AT of 0 msec, the wave V response was as fully masked as

possible (not discernible in the averaged response). The

"standard" recovery function was obtained by recording the

probe responses for the different ATs. Following the

"standard" series, the subject received one of the following


1. Masker duration was changed to 30 and then 120
msec and responses were recorded for each AT.

2. Masker level was changed -10 dB and then +10 dB
relative to the "standard" series and responses
were recorded for each AT.

3. The low frequency cutoff of the high-passed
masker (60 msec duration) was systematically
raised while maintaining a constant spectral
level set by the "standard" series. For each
of the makers 2-10, 4-10, and 6-10 kHz,
responses were recorded for each AT.


Auditory Brainstem Responses (ABRs) were collected on

a group of 15 normal hearing adults. In addition, whole-

nerve action potentials (APs) were simultaneously recorded

from some of the subjects. Using a forward masking paradigm,

responses were recorded only for the probe click. Not all

recording sessions produced usable AP data because of either

an inability to obtain proper placement of the ear canal

electrode after repeated attempts or the subject decided

against the use of the ear canal electrode. Out of 17 re-

cording sessions where an ear canal electrode was attempted,

9 produced usable AP data. One reason for difficulty with the

ear canal electrode was that the wrapped end often did not

make contact with the canal wall, e.g., the tubing would

"hang up" on a protruding edge of the canal or cerumen. In

addition, if the electrode was not positioned properly on

the first attempt it was difficult to reposition the electrode

(because it sprang open) without causing discomfort to the

subject. If the electrode could not be repositioned easily,

it was removed and only the ABR was collected.


Each subject received the "standard" series; 60 msec

wide-band masker set at an intensity level sufficient to

eliminate as completely as possible (fully mask) the wave V

peak of the ABR when .LT=0 msec. Responses to the probe

click were then obtained for each AT. All subjects received

the "standard" series which provide the largest amount of

data and which served as a reference for subsequent variation

of masker parameters. Three subjects (DR, TL, PD) received

two of the additional series, i.e., variation in masker dura-

tion, level, or low frequency cutoff in the same (extra long)

session and two subjects (DS, MO) were used for two of the

additional series on separate days. Table I summarizes the

recording series for each subject and includes whether or not

corresponding ear canal responses were obtained. ABR data as

a function of AT were obtained for 7 subjects for each of the

variations in masker parameter and AP data was obtained for

at least three subjects for each variation in masker.

Examples of the ABR and AP waveforms to the probe click

(50 dB HL) without the presence of the forward masker, i.e.,

unmasked control responses, are shown in Fig. 3. In the ABR

waveforms on the left of Fig. 3 the wave V peak is easily

identifiable in all of the examples. Peaks earlier than

wave V in the ABR can also be seen in some of the control

Table I. Forward masking series received by each of the
subjects (+) indicates success and (-)
indicates failure in obtaining data for ABR/AP.

Masking Condition
High-Pass Masking Masking "Standard"
Subject Masking Duration Level Series

KC +/+ +/+
DS +/+ +/- +/+
DR +/+ +/+ +/+
MO +/+ +/- +/+
TL +/+ +/+ +/+
PD +/- +/- +/-
NK +/- +/- +/-
BK +/+ +/+
KG +/- +/-
MS +/+ +/+
CB +/- +/-
SP +/+ +/+
SL +/- +/-
SW +/+ +/+
GC +/- +/-

Total 7/5 7/3 7/3 15/9

0) r0
4 ra

01 0
Z -1-


> (-H


Q) 4-4
> *r


4-4 Z

0 ,

4-t C





0 o


u -r-I
-- r-i

C 04
-0 0

-.1( 0




< c/

responses, but are not as easily identifiable in all sub-

jects. Only the wave V peak was systematically measured.

The latencies were measured to the positive wave V peak

just prior to the characteristic sharp negative trough

occurring in the appropriate latency range for this response,

i.e., between 6 and 10 msec. In some instances wave V

appears as a shelf on the positive peak just before the

negative trough, an example of which can be seen for subject

KC in Fig. 3. The peak-to-peak wave V amplitudes were

measured from the wave V peak to the following negative

trough. For the 15 subjects, the mean latency and t 1

standard deviation (S.D.) for the unmasked control wave V is

6.815 .220 msec. The mean peak-to-peak amplitude of the

control wave V is .223 + .067. volts. The AP responses from

the ear canal, shown on the right portion of Fig. 3, corre-

spond to the simultaneously obtained ABRs shown on the left.

The AP shows an initial negative trough (Nl) followed by a

larger positive peak which is generally broader in time than

the N1 trough and has a negative-going dip superimposed on

the positive peak which never goes below baseline in any

of the subjects. There is also considerable variability in

the configuration of the ear canal AP across subjects. Those

subjects with smaller amplitudes have a broader waveform

and a less noticeable negative-going dip on the positive

peak, most likely a result of the relation of the electrode

position to the potential sources. Latency measurements of

the AP are made to the base of the N1 trough. For the

unmasked N1 controls, the mean latency is 2.864 .135 msec.

The peak-to-peak amplitudes are measured from the N1 trough

to the nearest following positive peak. The mean N1 control

amplitude is .461 .200 volts. Considerable intersubject

variability of N1 amplitude was apparent and was primarily

related to the position of the ear canal electrode. Intra-

subject N1 amplitude variability across the recording session

was relatively low. The AP waveforms are similar to those

found by other investigators from the ear canal (Coats, 1964a,

1976; Yoshie, 1968) except for the dominance of the positive

peak following N1 and a poorly defined N2 in the AP waveforms

of this investigation.

Comparisons between the unmasked control and the forward

masked responses are made as a function of the time interval

between masker termination and probe onset (AT) for both

the mean latencasand mean amplitude of wave V and Ni. The

recovery function, defined as the probe response magnitude

(latency or amplitude) as a function of AT, represents the

time course of recovery from forward masking and is described

below for the different masker variables.

"Standard" Recovery Function

The "standard" forward masking series was obtained for

the ABR in all 15 of the subjects and AP data were included

from 9 of the subjects. Only one "standard" series for each

subject was included in the analysis. Masker levels necessary

to "fully mask" wave V (AT=0 msec) ranged from 74 to 82 dB

SPL. Preliminary data on two subjects using a simultaneous

continuous wide-band noise masker resulted in about 8-9 dB

less masker level necessary to fully mask wave V than for

the pulsed (60 msec) masker at AT=0 msec. Leaving the

masker level constant and switching from continuous to pulsed

in the simultaneous presentation maintained the fully masked

condition. The levels of masker necessary to fully mask wave

V were considerably higher than needed to fully mask the N1

response. In fact, it was often difficult to be absolutely

certain when wave V was fully masked because of a seemingly

apparent resistance to masking, especially in a forward mask-

ing paradigm, or because of a relatively more "noisy" base-

line. As the masking noise was added to the recording

paradigm and the intensity raised, the most apparent effect

on wave V was an increase in latency. With additional in-

creases in masker level, the amplitude of wave V diminished.

In some cases, adding 2 or 3 dB of masker caused the wave V

peak to go from almost full amplitude (with longer latency)

to unobservable.

Figure 4 shows typical unmasked controls and forward

masked ABRs along with corresponding AP waveforms for the

"standard" series. The probe-evoked ABRs and APs show

alterations produced by the preceding wide-band masker.

Wave V is easily observed in the masked waveforms and shows

a considerable increase in latency (relative to control)

for short ATs. Wave V amplitude is not severely altered

except at the earliest ATs. In addition, the peaks earlier

than wave V are not clearly identifiable in the masked wave-

forms. (Only 1200-1500 samples were used for the averaged

responses and adding more samples may have improved the clarity

of the earlier peaks.) The N1 latency, on the other hand,

shows only a slight latency shift even for the shortest AT,

but the N1 amplitude is considerably reduced even for the

longest AT.

The "standard" latency recovery function based on the

mean values are plotted in Fig. 5 for wave V, N1, and the

computed latency difference between wave V and N1 (V-N1)

representing the central brainstem conduction time. The

unmasked control (CNTL) values are also included in Fig. 5

and subsequent figures. For wave V (top of Fig. 5), the

E-1 ( >

o 2 -H
-4 E
-W -0
uO -4
O a)
fl >i

2 020O
4 rd 4-)
0 e

-H -H
(0 1
a) -


04 -1 4-)

SC (02

4U) 0

a) U) a)
Q r s-

44 a

0 E-
04 (

5, ro 0


< <<




1 T


/,9 1 ".L .... . . .....
3.5 r

0 6 12 25 58 100 CNTL


Fig. 5. Mean and 1 S.D. of wave V, Nl, and V-N1
latencies as a function of AT (delay time)
for the "standard" series. Unmasked con-
trols (CNTL) are indicated on the right in
this and subsequent figures.

maximum mean latency shift (increase) occurs at AT=6.2

msec and is 1.288 msec longer than the control latency.

For AT=0 msec, there are no wave V responses because of

the criterion (fully masked. As AT increases, the mean

latency of wave V systematically decreases towards the

control latency. The time course of wave V latency recovery

from forward masking can be described as a linear function

of log AT between the values of AT used in this investi-

gation, i.e., 6.2-100 msec. Although AT values above and

below 100 and 6.2 msec were not included, one may expect

deviations from the logrithmic function for other values of

AT. However, between ATs of 6.2 and 100 msec the equation

LATV = -.92 log AT + 8.867 fits the mean wave V latency

data quite well. For all ATs, the mean masked wave V

latency is greater than the mean control wave V latency.

For the "standard" series, as well as for each of the series

described in following sections for the different masker

variables, an analysis of the data was performed using

Statistical Analysis System (SAS) programs (University of

Florida Computer Center). An analysis of variance appro-

priate for unbalanced data (General Linear Systems, GLM)

was used and included partitioning out the variance due to

subjects as well as the different "treatment" variables. As

expected, for wave V latency for the "standard" series,LT

was found to be highly significant (F=191.24, df=5, 69

p < .001. The unmasked controls were not included in any of

the analyses of variance. In addition, whenever the ampli-

tudes of the responses were not measureable they were assigned

a value of 0 uvolts, but their corresponding latencies were

treated as "missing" in the analyses. An additional analysis

was performed in order to test which of the "treatment" means

are significantly different (p < .05) from the unmasked con-

trol mean, i.e., to determine where complete recovery occurs.

This was done using a paired-comparison procedure based on

least-square means (for unbalanced data) by SAS (LSMEANS/

PDIFF) for each possible comparison. (A Duncan Multiple

Range Test was also included to test for differences among

means (p < .05) and the results were in almost all cases

the same as the results for the LSMEANS test). Table II

gives the means and standard deviations of wave V latency

as a function of L T for the "standard" series. Also in-

cluded in Table II and subsequent tables are indicators

(*) for those masked response means which are significantly

different (p < .05) from the unmasked control response mean.

As Table II shows, wave V latency is significantly different

from control even for-LT=100 msec, i.e., complete recovery

has not occurred.

14 n o
E- r-l N
Z in co N
U *r

4-s1 >i

m cr 41

r,- rd

,- --, -I
W 3Q 4-4

4 C r


a ) :: E-4
4 *< *

> ~m

>0 >->
a 0
o4- rd d (d

uo o
CO -i

0-.i r J
0 4 .0

41 $.1 *.I.
0 4- m d

-> -H

-i s 4J

-1 0 l U.
t o

r0 r mC 0
4 4-)
cm 4 '-

U ct- s

a4 a1 4-4




o c.o




() CN




0 0o


*' L

LA 7






dn >
~r -4



S a



r-I r-i
r-4 ,-


a a

The N1 "standard" latency recovery function is shown

in the bottom portion of Fig. 5. As for wave V, the maximum

mean latency shift of N1 occurs at the shortest AT; however,

the amount of maximum N1 latency shift, .444 msec, is only

about one-third that seen for wave V (1.288 msec). The time

course of the N1 latency function, like for wave V, can

also be described as a linear function of log AT between ATs

of 6.2 and 100 msec, but the slope of the N1 recovery func-

tion is much less steep than for wave V. The mean N1 latency

data as a function of AT fit the equation, LATN1 = -.16

log AT + 3.433. For NI, AT is also highly significant

(F=14.44, df=5, 35 p < .001). Table II includes the means

and corresponding statistics for N1 latency as a function

of A T. N1 latency has completely recovered by AT=50 msec.

The V-N1 latency interval (middle portion of Fig. 5)

also changes as a function of AT as expected due to the

longer latency changes of wave V relative to NI. The mean

unmasked V-N1 latency interval is 3.956 t .303 msec and is in

agreement with other investigators (Davis, 1976; Elberling,

1976; Rowe, 1978; Coats et al., 1979). Again, AT is highly

significant (F=14.56, df=5, 35 p< .001). The V-N1 latency

interval is equal to the control latency for ATs of 50 and

100 msec (see Table II).

The amplitude recovery functions, in uvolts, for wave

V and N1 are shown in Fig. 6. Although the variability is

large, there is a systematic increase in amplitude for both

wave V and N1 as -T increases. However, the masked wave V

amplitude is less than the wave V control amplitude only

for ATs below 25 msec, while the masked N1 amplitude is

dramatically reduced at all ATs. To describe the amplitude

recovery functions in another way,the mean amplitudes are

normalized by considering the amplitude of the mean unmasked

response as 100% and the mean amplitudes for the responses

at each LT are then expressed as a percent of the mean

unmasked response (% control amplitude). The percent con-

trol amplitude is used to describe the remainder of the

amplitude recovery functions. Figure 7 shows the amplitude

recovery functions for both wave V and N1 and one can easily

notice the different behavior of the two responses.

For wave V, the amplitude recovery function does not

follow a simple logrithmic function as was seen for the wave

V (and N1) latency recovery functions; however, a monotonic

growth function is apparent across AT. For wave V ampli-

tude, LT was a significant factor (F=8.72, df=5, 69 p < .001).

Wave V amplitude recovers from fully masked to 60% of con-

trol by AT=6.2 msec and has completely recovered (p < .05)

Fig. 6. Mean and l1 S.D. for wave V and N1 amplitude
voltst) as a function of AT for the
"standard" series.


-* K 3. 39-

^ I T 1 i
Iii 9.29


1 2<

a" 8.39

1 9.29 N


S. . 29 .L. ...

8 6 12 25 50 109


j 88.8

0 O

I / L

8 6 12 25 5o 1o8


Fig. 7. Mean wave V and N1 amplitude recovery
functions expressed as percent of the
mean unmasked control amplitude (1000%)
for the "standard" series.

by LT=25 msec (even though wave V latency is not recovered

even by ;T=100 msec). The slight supernormality of wave V

amplitude seen in Fig. 7 at LT=100 msec is not significantly

different from the control response. Table III gives the

means and corresponding statistics, as a function of 3T,

for wave V amplitude.

The effect of the forward makers on the amplitudes of

the N1 response was much greater than for wave V amplitudes.

As seen in Fig. 7, the mean amplitude (% control amplitude)

of N1 increases from fully masked to only 18% of the unmasked

control response at LT=6.2 msec and is only 65% of the control

by LT=100 msec. As expected, LT is a highly significant

factor for N1 amplitude (F=20.07, df=5, 37 p < .001). The mean

N1 amplitudes are significantly different from the control

response amplitude for all LTs, i.e., full recovery does

not occur for N1 amplitude by AT=100 msec (even though the

mean N1 latency is equal to control by AT=50 msec). The

N1 amplitude recovery function is less steep than the wave V

amplitude recovery function. Table III gives the summary

data for N1 amplitude along with the wave V data.

Figure 8 shows a summary comparison of wave V and N1

latency and amplitude behavior for the "standard" series

from which the following characteristics can be summarized:

n (CN

Ln It
in CNt

-LA (N
.-4 (N

LA- (N




zr 00
i- '-


ON m
0"1 (N



r0 0

cN 00




Z a,

- I

b -.
a a a~ P e
C1O NOD.L.Cfla3i 2CfL.ldWV





II 5


U -


m0 w

0l (
4 c

m un

0 Q0

4 J
p M
0 0,

4- -I
4 a


a -

U) ()


-4 0

E 4
0 a


) -m

0 41
0) -4

E E (d
0 (
> >

U m
*H X) 0

-- -O

0 0C-
O> -I

4J 4.



Us E

ra 0 0 .
E 0
CO ( l
0 U IIr

0 >*

, O ,-
(0 0 0

.- u -4
-4 CU >-I
(U (U (

-- ~I


1. As AT increases, the latencies and amplitudes
of both wave V and N1 monotonically approach
the control response.

2. The most pronounced effects of the wide-band for-
ward masker are a decrease in the N1 amplitude
and an increase in the wave V latency.

3. The latency of wave V is not completely recovered
to the control response latency by AT=100 msec;
however, the amplitude of wave V is completely
recovered by AT=25 msec.

4. The latency of N1 is completely recovered by
AT=50 msec, but N1 amplitude is not recovered
even by AT=100 msec.

The Effect of Masker Duration

The effects of changing the duration of the wide-band

forward masker on the probe-evoked wave V response were investi-

gated in 7 subjects and comparable N1 data were obtained for 3

of those subjects. The mean values of the probe response

latency and amplitude were computed for each AT for each of

the three masker durations (30, 60, 120 msec). Recall that

changing the masker duration resulted in a change in the

intermasker interval (IMI) since the repetition rate remained

constant. In two of the subjects it was observed that there

were no noticeable differences in the wave V responses between

constant IMI (rate changes) and constant rate (IMI changes).

The effect of keeping IMI constant vs keeping rate constant

was not evaluated for the N1 response.

Figure 9 shows the mean latency recovery functions for

wave V and N1 for masker durations of 30, 60 ("standard"),

and 120 msec. For wave V there is a systematic increase in

the probe response latency as the masker duration increases

and nearly equal shifts of latency are observed at each T.

There is no significant interaction between masker duration

and ~T. Masker duration is a significant factor for wave V

latency (F=24.68, df=2, 85 p <.001); however, only the mean

responses to the 120 msec masker are significantly different

from the responses to the 30 and 60 msec makers. The time for

complete recovery of wave V latency increases with increasing

masker duration. The mean wave V latency is fully recovered

by AT=50 msec following the 30 msec masker, by LT=100 msec

following the 60 msec masker (for this smaller subgroup of

subjects this "standard" series is recovered by LT=100

msec, whereas for the larger group of subjects described

earlier wave V latency had not recovered by AT=100 msec),

and is not fully recovered by AT=100 msec following the 120

msec masker. Table IV gives the summary data for wave V

latency for the different masker durations.

In contrast to the mean wave V latency, the mean N1

latency does not show any significant difference as a function

of masker duration (F=0.00, df=2) and the mean N1 latency


o 8.58


1- 4.S5"



a 3 weC
0 68 uc
+128 asc


L L/ L L k L.L L h .L I i- I
0 6 12 2Z 58 108 CNTL


Fig. 9. Latency recovery functions for wave V, N1,
and V-N1 probe responses following wide-band
makers with durations of 30, 60, and 120
msec. Level of the masker was set to "fully"
eliminate wave V response at AT=0 following
the 60 msec masker ("standard" series).


*I *
o ,-




r I


0 a






o d


0 e


.-4 )
a a

0 00
o Q 0

H *
a m

m r-
r4 M,

a a a










D N^ Mn LD CN m I' CN
40 0 010 0 0


CO N CN z ":T
OO 0) No co ,
SE- 00 co O r-4 r-I r-

C oM m o o

*-I . . .
u c n C

I co 0N o 0n -

N U r-I 0

0m 0 M O
Lo oN ro N

1 co 0 r00 0 m
t .

i ON m CD
E- in C CM Ni c n M l
MJ * * *


0 I

> a)
I If I f

r- > 0 0
z0 0 0 Q
o 0 o 0

C) 0 0 00 0 0



(U I

&Hl > ~ a >

recovery curves are the same for the responses following

all three duration makers (see Fig. 9). For all three

masker durations, it can be observed that the mean N1 probe

response latency increases slightly at the shorter ATs, but

the amount of N1 latency shift is much less than the wave V

latency shift. The mean N1 latency for this group of sub-

jects is fully recovered by.T=25 msec following all three

masker durations. As can be seen in the middle portion of

Fig. 9, the V-N1 latency interval increases with increasing

masker duration; however, the differences due to masker

duration are not significant (F=1.67, df=2). Table IV in-

cludes the summary data for N1 and V-N1 latencies for the

different masker durations.

The mean amplitude recovery functions for wave V are

shown in Fig. 10. For wave V amplitude, there is no signi-

ficant interaction between masker duration and!T on the

probe-evoked responses and the effect of masker duration is

significant (F=11.65, df=2, 99 p < .001), but only the mean

responses following the 30 msec masker are significantly

different from the responses following the 60 and 120 msec

makers. This is not consistent with differences found for

wave V latency, i.e., the mean response following the 120

msec masker was different from the mean response for the

A 39 moe
0 68 mec
S128 9s*c




I - I I I I 1 I 1 | r |

6 12

25 58


Fig. 10.

Amplitude recovery functions for wave V probe
responses following wide-band makers with
durations of 30, 60, and 100 msec. Level
of the masker was set to"fully" eliminate
wave V response at AT=0 msec for the 60
msec masker.


30 and 60 msec makers. The recovery of the wave V ampli-

tude is faster following the 30 msec masker than following

the 60 and 120 msec makers which show similar recovery curves.

An interesting feature seen in Fig. 10 is an apparent super-

normality of the mean wave V amplitude forATs above 25

msec and this supernormality is most prevalent for the 30

msec masked condition. However, none of the supernormal

amplitudes are significantly different from the control

response. Wave V amplitude is equal to the control response

amplitude for all ZTs above 12.5 msec for the 60 and 120

msec masked conditions, and for ATs above 6.2 msec for the

30 msec masked condition. Table V gives the summary data

for wave V amplitude for the three masker durations.

The mean amplitude recovery functions for N1 for all

three masker durations are shown in Fig. 11. Increasing

masker duration results in a decrease in the postmasking

amplitude of the N1 probe response. For Ts below 25 msec,

the N1 responses were difficult to measure in all three

subjects for the 120 msec masked condition (and wave V was

easily observed). Unlike N1 latency, which showed no

significant duration effect, N1 amplitude is significantly

affected by masker duration (F=7.79, df=2, 33 p <.001) with

the difference occurring between the mean responses to the

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - Version 2.9.9 - mvs