J Am Acad Audiol 20:597–606 (2009)
Amplitude Modulation Detection by Listeners with
Unilateral Dead Regions
DOI: 10.3766/jaaa.20.10.2
Brian C.J. Moore*
Vinay{
Abstract
Background: A dead region is a region in the cochlea where the inner hair cells and/or neurons are functioning very poorly, if at all. We have shown that, for people with sensorineural hearing loss, thresholds for
detecting sinusoidal amplitude modulation (AM) of a sinusoidal carrier were lower for ears with high-frequency
dead regions, as diagnosed using the threshold-equalizing noise test, calibrated in hearing level, than for ears
without dead regions when the carrier frequency was below the edge frequency, fe, of the dead region.
Purpose: To measure AM-detection thresholds for subjects with unilateral dead regions, using carrier
frequencies both below and above fe.
Research Design: Ten subjects with bilateral high-frequency hearing loss, but with unilateral highfrequency dead regions, were tested. The carriers were presented at sensation levels of 5, 10, or
15 dB. The values of fe were close to 1000, 1500, or 2000 Hz.
Results: For carrier frequencies below fe, AM-detection thresholds were lower for the ears with dead regions
than for the ears without dead regions, replicating earlier findings. In contrast, for carrier frequencies above fe,
AM-detection thresholds tended to be higher for ears with dead regions than for ears without dead regions.
Conclusions: The reason why AM detection was poorer in the ears with dead regions for carrier frequencies above fe is unclear. However, this finding is consistent with the generally poor discrimination
of sounds that has been reported previously for sounds with frequency components falling within a
dead region. The results have implications for the ability of people with dead regions to use information
from frequency components falling inside the dead region.
Key Words: Amplitude modulation detection, cortical plasticity, dead region
Abbreviations: 2AFC 5 two-alternative forced choice; AM 5 amplitude modulation; ERBN 5 equivalent
rectangular bandwidth of the auditory filter averaged across young listeners with normal hearing;
fe 5 edge frequency of dead region; IHC 5 inner hair cell; m 5 modulation index; TEN 5 thresholdequalizing noise; TEN(HL) 5 threshold-equalizing noise, calibrated in hearing level
A
dead region is a region in the cochlea where the
inner hair cells (IHCs) and/or neurons are
functioning very poorly, if at all (Moore, 2001,
2004). Dead regions are known to occur in humans and
may be caused by a variety of factors, including exposure
to impact sounds and infections (Schuknecht, 1993;
Halpin et al, 1994; Borg et al, 1995). Little or no
information is transmitted to the brain about basilar
membrane vibration in a dead region. However, a tone or
other narrowband signal producing peak vibration in that
region may be detected by off-place (off-frequency)
listening. For example, if there is a dead region at the
base of the cochlea, which normally responds to high
frequencies, a high-frequency tone may be detected, if it is
sufficiently intense, via the spread of basilar membrane
vibration to a region tuned to lower frequencies. This is
the basis for psychophysical tests for detecting a dead
region, such as psychophysical tuning curves (Thornton
and Abbas, 1980; Florentine and Houtsma, 1983; Moore
and Alcántara, 2001; Kluk and Moore, 2005, 2006b; Sek et
al, 2005) and the threshold-equalizing noise (TEN) test
(Moore et al, 2000; Moore et al, 2004).
The TEN test involves measuring the threshold for
detecting a pure tone presented in a background noise
*Department of Experimental Psychology, University of Cambridge; {All India Institute of Speech and Hearing, Manasagangothri, Mysore
Brian C.J. Moore, Department of Experimental Psychology, University of Cambridge, Downing Street, Cambridge, CB2 3EB, England;
E-mail: [email protected]; Phone: +44 1223 333574; Fax: +44 1223 333564
597
Journal of the American Academy of Audiology/Volume 20, Number 10, 2009
called ‘‘threshold-equalizing noise.’’ The noise spectrum
is shaped in such a way that, for people with normal
hearing, the threshold for detecting a pure tone in the
noise is almost constant over a wide range of tone
frequencies. The masked threshold is approximately
equal to the nominal level of the noise, specified as the
level in a 132 Hz wide band centered at 1000 Hz. The
value of 132 Hz corresponds to the equivalent rectangular bandwidth of the auditory filter, as determined using
young, normally hearing listeners, which is denoted
ERBN (Moore, 2003). When the pure-tone signal frequency falls in a dead region, the signal will only be
detected when it produces sufficient basilar membrane
vibration at a remote region in the cochlea where there
are surviving IHCs and neurons. The amount of
vibration at this remote region will be less than in the
dead region, and so the noise will be very effective in
masking it. Thus, the signal threshold is expected to be
markedly higher than normal. A masked threshold that
is 10 dB or more higher than normal is taken as
indicating a dead region (Moore et al, 2000; Moore et
al, 2004). The extent of a dead region is defined in terms
of its edge frequency (or frequencies), fe, which corresponds to the characteristic frequency of the IHCs and/or
neurons immediately adjacent to the dead region (Moore
et al, 2000). A dead region can be patchy, occurring over
one or more restricted regions in the cochlea (Moore and
Alcántara, 2001). Here, the term high-frequency dead
region is used to refer to a dead region that starts at fe
and extends upward from there.
Studies using animals have shown that damage to the
basal region of the cochlea, effectively producing a highfrequency dead region, can lead to the region of the
auditory cortex that previously responded to the damaged region of the cochlea becoming ‘‘tuned’’ to frequencies corresponding to the adjacent relatively undamaged
region (Robertson and Irvine, 1989). The possible
functional consequences of this have been explored in
several studies; for reviews, see Thai-Van and colleagues
(2007) and Moore and Vinay (2009). It has been shown
that, for people with high-frequency dead regions,
frequency discrimination is enhanced for frequencies
that fall just below fe (McDermott et al, 1998; Thai-Van
et al, 2003; Kluk and Moore, 2006a; Moore and Vinay,
2009). This has generally been interpreted in terms of
cortical plasticity; the expanded cortical representation
of frequencies just below fe is assumed to be responsible
for the enhanced ability to discriminate those frequencies
(McDermott et al, 1998; Thai-Van et al, 2007).
Recently, we (Moore and Vinay, 2009) showed that
the ability to detect amplitude modulation (AM) was
also affected by the presence of a dead region. We tested
two groups of subjects, with and without acquired highfrequency dead regions, as assessed using the TEN test
calibrated in hearing level, or TEN(HL) (Moore et al,
2004). All subjects in both groups had high-frequency
598
bilateral hearing loss. For the ears with dead regions,
the value of fe was close to 1000 or 1500 Hz. The two
groups were matched in terms of audiometric thresholds for frequencies below fe and in terms of age. Three
subjects with unilateral dead regions (with matched
low-frequency audiometric thresholds across ears) were
also tested. Thresholds for detecting sinusoidal AM of a
sinusoidal carrier were measured. Carrier frequencies
of 500 and 800 Hz were used with all subjects, and an
additional carrier frequency of 1200 Hz was used for
ears with fe close to 1500 Hz and their matched
counterparts. The carrier level was 20 dB SL. Modulation frequencies were 4, 50, and 100 Hz. AM-detection
thresholds were generally lower (better) for the ears
with dead regions than for the ears without dead
regions, and this effect was statistically significant.
For the subjects with unilateral dead regions, thresholds were lower for the ears with dead regions than for
the ears without dead regions. The ears with dead
regions showed better performance than the ears
without dead regions even for a carrier frequency
(500 Hz) that was well below fe, and this effect was
roughly independent of modulation frequency.
One problem in interpreting the improved AMdetection thresholds is that cochlear hearing loss and
the associated loudness recruitment can lead to improved
AM detection at low SLs; for a review, see Moore (2007).
Indeed, this is the basis for the short-increment
sensitivity index test (Jerger, 1962; Buus et al, 1982a,
1982b). We have argued that loudness recruitment was
unlikely to have influenced the difference between
results for the ears with and without dead regions, since
the ears were matched in terms of low-frequency
audiometric thresholds, and the amount of loudness
recruitment caused by cochlear hearing loss is closely
related to the amount of hearing loss (Miskolczy-Fodor,
1960; Moore et al, 1999; Moore and Glasberg, 2004). Also,
some of the subjects that we tested had normal hearing
for frequencies of 1000 Hz and below, but, for those
subjects, AM-detection thresholds were lower for the ear
with a dead region than for the ear without a dead region.
However, it is possible that cortical reorganization might
itself lead to a form of loudness recruitment. The cortical
overrepresentation of the low-frequency region of the
cochlea that occurs following a high-frequency dead
region might lead to a more-rapid-than-normal growth of
loudness with increasing sound level, and that in itself
might lead to improved AM detection.
We (Moore and Vinay, 2009) did not assess AM
detection for carrier frequencies that fell within a dead
region, that is, fell above fe. There is evidence that
loudness recruitment is very marked for tones that fall
within a dead region (McDermott et al, 1998; Thai-Van
et al, 2003). If the enhanced AM detection we found for
ears with dead regions was a consequence of greater
loudness recruitment in the ears with dead regions, then
Amplitude Modulation Detection/Moore and Vinay
one would expect enhanced AM detection for tones
falling above fe, as well as for tones falling below fe.
Another consideration is that tones with frequencies
above fe are detected via off-frequency or off-place
listening; they are detected at a place in the cochlea
tuned to frequencies just below fe. Since those frequencies are expected to be overrepresented in the cortex, one
might again expect enhanced AM detection for carriers
that fall above fe. On the other hand, there is evidence
that pure tones with frequencies above fe are often heard
as sounding noise-like (Huss and Moore, 2005a) and do
not have a clear pitch (Huss and Moore, 2005b). Reports
of a noise-like sensation for an unmodulated tone might
indicate that the tone sounds as if it is fluctuating to
some extent. This might make AM detection more
difficult (Stone et al, 2008b), so AM detection might be
worse than for an ear without a dead region. In the
present study, we assessed how the presence of a dead
region affects AM detection by comparing results for the
two ears of subjects with bilateral hearing loss but
unilateral high-frequency dead regions, including carrier frequencies both below and above fe.
The results for frequencies above fe are of clinical
relevance since the processing of AM is thought to be
important for speech intelligibility (Drullman et al,
1994a, 1994b; Shannon et al, 1995). If the ability to
detect AM is poorer for frequencies above than below
fe, this would imply a poorer ability to process AM for
frequencies above fe, which might be associated with a
reduced ability to process AM in speech.
normal middle ear function, as assessed using a GrasonStadler Tympstar Immittance meter and indicated by
the lack of air–bone gaps greater than 10 dB. Transient
evoked otoacoustic emissions were assessed using click
stimuli and an Otodynamics ILO 292 analyzer and were
found to be absent for all subjects. The ages of the
subjects ranged from 37 to 78 years. There were seven
males and three females. All subjects had postlingually
acquired hearing loss, and all had a hearing loss for five
years or more at the time of testing. None of the subjects
had nonauditory neural conditions, and none had any
history of ear discharge or a feeling of fullness in the ear.
Each ear of each subject was considered separately; the
right ear is denoted RE, and the left ear is denoted LE.
The diagnosis of dead regions was based on the results of
the TEN(HL) test (Moore et al, 2004); see below for
details. S1RE, S2LE, S3RE, S4LE, S5LE, S6LE, S7RE,
S8RE, S9RE, and S10RE were diagnosed to have dead
regions. Generally, absolute thresholds were similar for
the two ears of each subject for audiometric frequencies
below fe, but audiometric thresholds tended to be higher
for the ears with dead regions for frequencies at and
above fe. Exceptions were S3 and S9, whose audiometric
thresholds were very similar for the two ears over the
whole range of audiometric frequencies.
For each subject, all testing was done in a single
session, with breaks as required. The typical duration of
a session was two hours, but this varied across subjects.
METHOD
After the initial diagnosis of sensorineural hearing
loss, the TEN(HL) test was conducted using the same
audiometer. The TEN(HL) test compact disk (Moore et
al, 2004) was replayed via a Philips 729 K CD player
connected to the audiometer. Test frequencies were
0.5, 0.75, 1, 1.5, 2, 3, and 4 kHz. The level of the signal
and the TEN(HL) were controlled using the attenuators in the audiometer. The TEN(HL) level was 70 dB
HL/ERBN. The signal level was varied in 2 dB steps to
determine the masked thresholds, as recommended by
Moore and colleagues (2004). A ‘‘no response’’ (NR) was
recorded when the subject did not indicate hearing the
signal at the maximum output level of the audiometer,
which was 100 dB HL for the signals derived from the
TEN(HL) CD. NR implies that the threshold was above
100 dB HL. The diagnosis of the presence or absence of
a dead region at a specific frequency was based on the
criteria suggested by Moore and colleagues (2004). If
the masked threshold in the TEN(HL) was 10 dB or
more above the TEN(HL) level/ERBN (i.e., was 80 dB
HL or more), and the TEN(HL) elevated the absolute
threshold by 10 dB or more, then a dead region was
assumed to be present at the signal frequency. If the
masked threshold in the TEN(HL) was less than 10 dB
above the TEN(HL) level/ERBN, and the TEN(HL)
Subject Selection and Audiometric Assessment
The subjects were selected from a large pool of
potential subjects who had been tested using the
TEN(HL) test (see below for details). Subjects generally
attended the clinic for the purpose of being fitted with a
hearing aid. This usually happened after an audiologic
evaluation, following clearance by a medically qualified
ear specialist. Ten subjects were chosen who had
sensorineural loss in both ears, with a dead region in
one ear and no dead region in the other, as diagnosed
using the TEN(HL) test. The purpose of the study was
explained to the subjects, and their consent was obtained
for participation in the study. The subjects are denoted
S1 to S10. Details of the subjects are given in Table 1.
Audiometric air-conduction thresholds were measured using a Madsen OB922 dual-channel clinical
audiometer with Telephonics TDH39 headphones, for
frequencies from 0.25 to 8 kHz. Bone-conduction thresholds were obtained for frequencies from 0.25 to 4 kHz
using a Radio Ear B71 bone vibrator. In both cases, the
modified Hughson-Westlake procedure described by
Carhart and Jerger (1959) was used. All subjects had
Application of Threshold-Equalizing Noise Test
599
Journal of the American Academy of Audiology/Volume 20, Number 10, 2009
Table 1. Audiometric Thresholds (Upper Entries in Each Row) and Thresholds in the Threshold-Equalizing Noise,
Calibrated in Hearing Level, Test (Lower Entries in Each Row) for Each Subject
Frequency (Hz)
Subject/Ear
S1RE
Age
Gender
250
500
750
1000
1500
2000
3000
4000
8000
76
M
35
45
70
55
70
55
72
60
76
55
70
60
74
40
70
45
70
55
72
60
76
30
70
40
70
65
72
45
70
45
72
30
70
60
74
60
74
40
70
40
74
50
70
50
76
55
72
60
74
65
76
70
78
45
70
55
70
65
74
65
74
35
70
55
70
60
76
45
70
45
74
30
70
70
76
65
74
45
76
40
70
50
70
50
74
55
70
65
74
70
76
70
76
45
70
55
74
65
74
70*
82
40
72
70
76
55
74
40
70
65*
82
35
70
75
78
70
78
50
72
45
72
60
76
55
74
65
74
70*
82
75*
86
70
78
50
72
60
72
70
76
75
86
45
72
75
78
65
74
45
70
70*
84
40
72
75*
86
75
78
75
78
50
74
70*
86
60
74
65
74
70
84
75
NR
75
78
50
74
75*
86
70
74
75
86
55
74
75*
86
70*
82
45
70
75
86
45
70
75
86
75
76
75*
86
60
76
75
NR
60
78
70
74
75
NR
75
NR
75
78
55
72
75
86
70
74
80
NR
55
72
80
NR
80
NR
55
74
70
84
55
70
75
NR
80
78
80
NR
70
74
75
NR
65
76
70
76
75
NR
80
NR
75
80
60
76
80
NR
80
78
80
NR
80
80
80
NR
80
NR
60
76
75
NR
65
72
80
NR
80
80
80
NR
75
78
100
S1LE
S2RE
45
70
M
S2LE
S3RE
65
66
M
S3LE
S4RE
78
M
77
M
37
F
76
M
57
F
S10LE
40
20
73
F
S9LE
S10RE
60
45
S8LE
S9RE
25
35
S7LE
S8RE
55
55
S6LE
S7RE
25
35
S5LE
S6RE
55
55
S4LE
S5RE
60
60
60
64
M
35
35
80
85
100
NR
100
85
NR
90
90
110
90
95
90
95
80
100
90
90
90
Note: Bold numbers indicate frequencies falling in a dead region; * indicates the edge frequency, fe; NR means no response.
elevated the absolute threshold by 10 dB or more, then
a dead region was assumed to be absent. It should be
noted that the ‘‘true’’ value of fe lies between the lowest
frequency at which the TEN(HL) test criteria were met
and the next lowest test frequency. Here, for simplicity, the value of fe was taken as the lowest frequency at
which the TEN(HL) test criteria for the presence of a
dead region were met.
Measurement of Absolute Thresholds
In addition to the measurement of audiometric
thresholds as described above, absolute thresholds for
600
pure-tone signals were measured with higher accuracy
in dB SPL using an adaptive two-alternative forcedchoice (2AFC) method. These absolute thresholds were
used to set the SLs of the stimuli used in the
measurement of AM detection. Stimuli were digitally
generated using a personal computer equipped with a
SoundMAX sound card. The sampling frequency was
25 kHz. The output of the sound card was fed to
Sennheiser HD265 earphones. These are ‘‘closed’’-type
earphones that give good isolation between the two ears.
They are diffuse-field equalized. The signal level was
started above the threshold level as estimated from the
audiogram. The signal could occur either in the first
Amplitude Modulation Detection/Moore and Vinay
interval or in the second interval, selected at random.
The signal lasted 200 msec, including 20 msec raisedcosine rise/fall times, and the intervals were separated
by 500 msec. The intervals were indicated by virtual
buttons on the computer screen (labeled 1 and 2), each of
which was lit up in blue during the appropriate interval.
The subject responded by clicking on the appropriate
button with a computer mouse or by pressing button 1 or
2 on the computer numeric keypad. Feedback was
provided by flashing the button in green for a correct
answer and red for an incorrect answer. A two-down,
one-up procedure was used (Levitt, 1971). Six turnpoints
were obtained. The step size was initially 6 dB. It was
changed to 4 dB after one turnpoint and to 2 dB after the
second turnpoint. The threshold was taken as the mean
signal level at the last four turnpoints.
Measurement of AM-Detection Thresholds
Thresholds for detecting AM were determined using an
adaptive 2AFC method, similar to that described for the
measurement of absolute thresholds, and using the same
equipment. One interval contained an unmodulated
carrier, and the other interval contained a carrier that
was amplitude modulated with modulation index m. The
order of the two was random. The task of the subject was to
indicate the interval in which the carrier was modulated.
The equation describing the modulated waveform as a
function of time, W(t), is W(t) 5 {1 + msin(2pfmodt +
Q)}sin(2pfcarrt), where fmod is the modulation frequency
and fcarr is the carrier frequency. A value of m 5 1 means
100 percent modulation, while a value of m 5 0 means no
modulation. The starting phase of the modulation, Q, was
random, and the AM was present throughout the whole
stimulus. The total power was equated across the two
intervals. The duration of each carrier was 1000 msec,
including 20 msec raised-cosine rise/fall times. The value
of m was adjusted using a two-down, one-up adaptive
procedure. The initial value was chosen to make the
modulation clearly audible. The value of m was adjusted
by a factor of 1.253 until two turnpoints had occurred.
Then m was adjusted by a factor of 1.252 until two more
turnpoints had occurred. Finally, m was adjusted by a
factor of 1.25 until eight further turnpoints had occurred.
The threshold was taken as the geometric mean of the
values of m at the last eight turnpoints.
Four carrier frequencies were used for each subject.
They were chosen so that, for the ear with a dead region,
two of the carrier frequencies fell below fe and two fell
above fe. The carrier frequencies were 500, 750, 2000,
and 4000 Hz for ears with fe 5 1000 Hz; 500, 1000, 2000,
and 4000 Hz for ears with fe 5 1500 Hz; and 500, 1000,
3000, and 4000 Hz for ears with fe 5 2000 Hz. For each
subject, the same carrier frequencies were used for the
ears with and without dead regions. The modulation
frequencies were 4, 16, and 50 Hz. These were chosen to
span the range of modulation frequencies that are
thought to be most important for speech perception
(Shannon et al, 1995; Stone et al, 2008a) while being low
enough to ensure that spectral sidebands produced by
the modulation were not resolved (Kohlrausch et al,
2000; Moore and Glasberg, 2001).
All subjects except one were tested using carrier levels
of 5, 10, and 15 dB SL. The exception was S5, who found
the stimuli to be too loud in both ears for the 2000 Hz
carrier frequency at 15 dB SL and who was tested using
levels of 5 and 10 dB SL only for that carrier frequency.
The order of testing the ear, SL, carrier, and modulation
frequencies was randomized for each subject. At least
two threshold estimates were obtained for each condition; where time permitted, an additional estimate was
obtained. The final threshold for a given condition was
taken as the mean across all estimates. In what follows,
thresholds are expressed as 20log10(m), that is, in
decibel-like units. For the maximum value of m (5 1),
the value of 20log10(m) is 0. The smaller the value of m,
the more negative 20log10(m) is. Thus, more negative
numbers indicate better performance.
RESULTS
T
he pattern of results was similar across subjects. To
simplify presentation, AM-detection thresholds were
averaged across subjects with the same value of fe.
Figure 1 shows mean AM-detection thresholds for S5
and S8, who both had fe 5 1000 Hz in one ear. Figure 2
shows mean thresholds for S2, S3, and S9, who all had fe 5
1500 Hz in one ear. Figure 3 shows mean thresholds for
S1, S4, S6, S7, and S10, who all had fe 5 2000 Hz in one
ear. In each figure, data for the ears with no dead region
are plotted on the left, and data for the ears with dead
regions are plotted on the right. For the ears with no dead
region, mean thresholds did not differ systematically for
carrier frequencies below the value of fe for the other ear
(filled symbols) and above that value (open symbols). In
contrast, for ears with dead regions, the mean thresholds
were consistently lower for carriers below fe than for
carriers above fe. Mean thresholds tended to improve
(decrease) with increasing SL, although this effect was
most apparent for carrier frequencies below fe presented
to the ears with dead regions. For carriers below fe, mean
thresholds were consistently lower for the ears with dead
regions than for the ears without dead regions. This
replicates our 2009 results. For carriers above fe, mean
thresholds tended to be higher for the ears with dead
regions than for the ears without dead regions.
The use of different carrier frequencies across
subjects made statistical analysis slightly awkward.
However, results were generally similar for the two
lower carrier frequencies (below fe) and the two higher
carrier frequencies (above fe) for each subject. Therefore, the data were analyzed in the following way:
601
Journal of the American Academy of Audiology/Volume 20, Number 10, 2009
Figure 1. Mean amplitude modulation detection thresholds, expressed as 20log10(m), for two subjects with a dead region in one ear with fe
5 1000 Hz. Data for ears with no dead regions are shown on the left, and data for ears with dead regions are shown on the right. The carrier
sensation level was 5 dB (top), 10 dB (middle), or 15 dB (bottom). Each symbol represents a different carrier frequency, as indicated in the
key. Filled symbols indicate carrier frequencies below fe, and open symbols indicate carrier frequencies above fe. Error bars show 6 one
standard deviation across subjects. Error bars are omitted when they would be smaller than the symbol used to represent the mean.
1. For each ear of each subject, and for each SL and
each modulation frequency, the AM-detection
thresholds, expressed as 20log10(m), were averaged
for the two lower carrier frequencies and the two
upper carrier frequencies. The resulting values are
denoted Av(low) and Av(high). Since S5 was not
tested using the carrier frequency of 2000 Hz at the
SL of 15 dB, the value of Av(high) at 15 dB SL was
based on the AM-detection threshold for the
4000 Hz carrier only.
602
2. A repeated-measures analysis of variance was
conducted on the values of Av(low) and Av(high),
with the following factors: ear (with dead region or
without dead region), carrier frequency region (below
or above fe), modulation frequency (4, 16, or 50 Hz),
and SL (5, 10, or 15 dB). The Greenhouse-Geisser
correction was used where appropriate.
The effect of ear was significant: F(1, 9) 5 7.44, p 5
.023. The mean threshold was slightly lower for ears
Amplitude Modulation Detection/Moore and Vinay
Figure 2. Mean amplitude modulation detection thresholds, expressed as 20log10(m), for three subjects with a dead region in one ear with
fe 5 1500 Hz. Data for ears with no dead regions are shown on the left, and data for ears with dead regions are shown on the right. The
carrier sensation level was 5 dB (top), 10 dB (middle), or 15 dB (bottom). Each symbol represents a different carrier frequency, as indicated
in the key. Filled symbols indicate carrier frequencies below fe, and open symbols indicate carrier frequencies above fe. Error bars show 6
one standard deviation across subjects. Error bars are omitted when they would be smaller than the symbol used to represent the mean.
with a dead region (216.3 dB) than for ears without a
dead region (215.1 dB). The effect of carrier frequency
region was significant: F(1, 9) 5 94.0, p , .001. The
mean threshold was lower for carriers below fe
(217.1 dB) than for carriers above fe (214.3 dB). The
effect of modulation frequency was not significant.
The effect of SL was significant: F(2, 18) 5 23.9, p ,
.001. The mean thresholds were 214.1, 216.2, and
216.8 dB for SLs of 5, 10, and 15 dB, respectively.
There was a significant interaction between ear and
carrier frequency: F(1, 9) 5 85.4, p , .001. Mean AMdetection thresholds were lower for the ears with dead
regions (219.1 dB) than for the ears without dead
regions (215.1 dB) for the carrier frequency region
below fe. A post hoc test (Tukey HSD) showed that this
difference was significant (p , .001). In contrast, for
the higher carrier frequency region, the mean threshold was higher for the ears with dead regions
(213.2 dB) than for the ears without dead regions
(215.1 dB). A post hoc test showed that this difference
was significant (p , .01). Other interactions either
were not significant or accounted for only a small
proportion of the variance in the data.
In summary, the results showed that, for carrier
frequencies below fe, AM-detection thresholds were
lower for ears with dead regions than for ears without
603
Journal of the American Academy of Audiology/Volume 20, Number 10, 2009
Figure 3. Mean amplitude modulation detection thresholds, expressed as 20log10(m), for five subjects with a dead region in one ear with fe
5 2000 Hz. Data for ears with no dead regions are shown on the left, and data for ears with dead regions are shown on the right. The carrier
sensation level was 5 dB (top), 10 dB (middle), or 15 dB (bottom). Each symbol represents a different carrier frequency, as indicated in the
key. Filled symbols indicate carrier frequencies below fe, and open symbols indicate carrier frequencies above fe. Error bars show 6 one
standard deviation across subjects. Error bars are omitted when they would be smaller than the symbol used to represent the mean.
dead regions. In contrast, for carrier frequencies above
fe, AM-detection thresholds tended to be higher for ears
with dead regions than for ears without dead regions.
DISCUSSION
T
he results indicate that AM-detection thresholds
obtained using sinusoidal carriers did not vary
markedly with AM frequency for frequencies up to
604
50 Hz. This is broadly consistent with previous results
for normally hearing subjects (Kohlrausch et al, 2000)
and hearing-impaired subjects (Moore and Glasberg,
2001), showing that AM-detection thresholds are roughly
independent of AM frequency for frequencies up to about
100 Hz, provided that the spectral sidebands produced
by the AM are not resolved in the auditory system.
However, it has sometimes been reported that AMdetection thresholds are lower for modulation frequencies
Amplitude Modulation Detection/Moore and Vinay
around 10 Hz than for frequencies around 80 Hz (Füllgrabe et al, 2003; Füllgrabe et al, 2005). The differences
across studies may be related to the carrier duration,
which was greater in the studies of Füllgrabe and
colleagues than in our study or the other studies cited
above. For normally hearing subjects, AM-detection
thresholds, expressed as 20log10(m), are typically about
215 dB for carriers at low SLs (Kohlrausch et al, 2000;
Moore and Glasberg, 2001; Füllgrabe et al, 2005),
although few data are available for SLs as low as those
used here. The typical threshold of 215 dB is comparable
to the AM-detection thresholds obtained here for ears
without dead regions. For the ears with dead regions, AMdetection thresholds appear to be somewhat better than
normal for carrier frequencies below fe but are comparable to or slightly worse than normal for carrier frequencies
above fe.
The finding that AM detection for carrier frequencies
below fe was better for ears with dead regions than for
ears without dead regions is consistent with the results
of our 2009 study. The present results show that this
occurred even when the carrier frequency was as much
as two octaves below fe (a carrier of 500 Hz with fe 5
2000 Hz). The present results also show that, for carrier
frequencies above fe, AM detection tended to be worse for
ears with than without dead regions.
There may be several reasons for the worse performance of ears with dead regions for carrier frequencies
above fe. It could be connected with the fact that, for
frequencies above fe, absolute thresholds were usually
higher for the ears with than without dead regions (see
Table 1). However, the absolute thresholds of S3 and
S9 were similar across ears, even for frequencies above
fe, and the pattern of their results is very similar to
that for the other subjects. Therefore, differences in
absolute threshold across ears do not seem to be
critical in accounting for the observed results.
Another possible factor accounting for the worse AM
detection of ears with dead regions for frequencies above
fe is the noise-like character of the percept that is usually
evoked by tones with such frequencies in ears with dead
regions (Huss and Moore, 2005a, 2005b). This noise-like
percept may indicate that even an unmodulated carrier
appears to fluctuate, which would make it harder to
detect AM imposed on a carrier. The noise-like percept
may result partly from a mismatch between place
information and the temporal information carried in
the patterns of phase locking in the auditory nerve. Such
a mismatch has been proposed previously to account for
the relatively poor frequency discrimination of tones
falling in a dead region (Huss and Moore, 2005a, 2005b;
Moore and Carlyon, 2005). The noise-like percept may
also occur because a modulated or unmodulated carrier
falling in a dead region is detected via the IHCs and/or
neurons immediately adjacent to the dead region, and
these may not be functioning in a normal way; rather,
there may be a poorly functioning region just below the
dead region (Huss and Moore, 2003). Consistent with
this view, for the ears with dead regions, the absolute
threshold for the audiometric frequency just below fe
was mostly in the range of 60 to 75 dB, indicating
hearing losses that were probably too large to be
explained by deficits in outer hair cell function alone
(Robles and Ruggero, 2001; Moore, 2007).
Whatever the reasons, the present results show that,
for an ear with a dead region, AM detection for sounds
with frequencies above fe is relatively poor. This may be
important because, as noted earlier, it is thought that
the perception of speech depends strongly on the
processing of patterns of AM in different frequency
bands (Drullman et al, 1994a, 1994b; Shannon et al,
1995). Indeed, since people with cochlear hearing loss
generally have reduced frequency selectivity (Glasberg
and Moore, 1986; Moore, 2007) and reduced sensitivity
to temporal fine structure in sounds (Hopkins and
Moore, 2007; Hopkins et al, 2008; Moore, 2008), they
may be especially dependent on the processing of
temporal envelope cues in order to understand speech.
Our results suggest that the processing of such cues is
likely to be less efficient for frequency components above
fe than for components below fe. This may partly explain
the limited benefit for speech intelligibility of amplifying
frequency components that fall well above fe (Vickers et
al, 2001; Baer et al, 2002; Preminger et al, 2005).
Acknowledgments. We thank the director of the All India
Institute of Speech and Hearing for providing the facilities to
carry out this work. The work of author BCJM was supported
by the Medical Research Council (U.K.). We thank Aleksander Sek for writing the software for conducting the amplitude
modulation detection task, Brian Glasberg for help with
statistical analyses, and Christian Füllgrabe for helpful
comments on an earlier version of this article. We also thank
three anonymous reviewers for helpful comments on an
earlier version of this article.
REFERENCES
Baer T, Moore BCJ, Kluk K. (2002) Effects of lowpass filtering on
the intelligibility of speech in noise for people with and without
dead regions at high frequencies. J Acoust Soc Am 112:1133–1144.
Borg E, Canlon B, Engström B. (1995) Noise-induced hearing loss—
Literature review and experiments in rabbits. Morphological and
electrophysiological features, exposure parameters and temporal
factors, variability and interactions. Scand Audiol 24 (Suppl 40):1–147.
Buus S, Florentine M, Redden RB. (1982a) The SISI test: a
review. Part I. Audiology 21:273–293.
Buus S, Florentine M, Redden RB. (1982b) The SISI test: a
review. Part II. Audiology 21:365–385.
Carhart R, Jerger JF. (1959) Preferred method for clinical
determination of pure-tone thresholds. J Speech Hear Disord
24:330–345.
Drullman R, Festen JM, Plomp R. (1994a) Effect of reducing slow
temporal modulations on speech reception. J Acoust Soc Am 95:
2670–2680.
605
Journal of the American Academy of Audiology/Volume 20, Number 10, 2009
Drullman R, Festen JM, Plomp R. (1994b) Effect of temporal envelope
smearing on speech reception. J Acoust Soc Am 95:1053–1064.
Florentine M, Houtsma AJM. (1983) Tuning curves and pitch
matches in a listener with a unilateral, low-frequency hearing
loss. J Acoust Soc Am 73:961–965.
Füllgrabe C, Meyer B, Lorenzi C. (2003) Effect of cochlear damage on
the detection of complex temporal envelopes. Hear Res 178:35–43.
Füllgrabe C, Moore BCJ, Demany L, et al. (2005) Modulation
masking produced by second-order modulators. J Acoust Soc Am
117:2158–2168.
Glasberg BR, Moore BCJ. (1986) Auditory filter shapes in
subjects with unilateral and bilateral cochlear impairments.
J Acoust Soc Am 79:1020–1033.
Halpin C, Thornton A, Hasso M. (1994) Low-frequency sensorineural loss: clinical evaluation and implications for hearing aid
fitting. Ear Hear 15:71–81.
Hopkins K, Moore BCJ. (2007) Moderate cochlear hearing loss
leads to a reduced ability to use temporal fine structure information. J Acoust Soc Am 122:1055–1068.
Moore BCJ. (2008) The role of temporal fine structure processing in
pitch perception, masking, and speech perception for normal-hearing
and hearing-impaired people. J Assoc Res Otolaryngol 9:399–406.
Moore BCJ, Alcántara JI. (2001) The use of psychophysical tuning
curves to explore dead regions in the cochlea. Ear Hear 22:268–278.
Moore BCJ, Carlyon RP. (2005) Perception of pitch by people with
cochlear hearing loss and by cochlear implant users. In: Plack CJ,
Oxenham AJ, Fay RR, Popper AN, eds. Pitch Perception. New
York: Springer, 234–277.
Moore BCJ, Glasberg BR. (2001) Temporal modulation transfer
functions obtained using sinusoidal carriers with normally hearing
and hearing-impaired listeners. J Acoust Soc Am 110:1067–1073.
Moore BCJ, Glasberg BR. (2004) A revised model of loudness
perception applied to cochlear hearing loss. Hear Res 188:70–88.
Moore BCJ, Vickers DA, Plack CJ, Oxenham AJ. (1999) Interrelationship between different psychoacoustic measures assumed
to be related to the cochlear active mechanism. J Acoust Soc Am
106:2761–2778.
Moore BCJ, Huss M, Vickers DA, et al. (2000) A test for the diagnosis
of dead regions in the cochlea. Br J Audiol 34:205–224.
Hopkins K, Moore BCJ, Stone MA. (2008) Effects of moderate
cochlear hearing loss on the ability to benefit from temporal
fine structure information in speech. J Acoust Soc Am 123:1140–1153.
Moore BCJ, Glasberg BR, Stone MA. (2004) New version of the
TEN test with calibrations in dB HL. Ear Hear 25:478–487.
Huss M, Moore BCJ. (2003) Tone decay for hearing-impaired
listeners with and without dead regions in the cochlea. J Acoust
Soc Am 114:3283–3294.
Moore BCJ, Vinay. (2009) Enhanced discrimination of lowfrequency sounds for subjects with high-frequency dead regions.
Brain 132:524–536.
Huss M, Moore BCJ. (2005a) Dead regions and noisiness of pure
tones. Int J Audiol 44:599–611.
Preminger JE, Carpenter R, Ziegler CH. (2005) A clinical
perspective on cochlear dead regions: intelligibility of speech and
subjective hearing aid benefit. J Am Acad Audiol 16:600–613.
Huss M, Moore BCJ. (2005b) Dead regions and pitch perception.
J Acoust Soc Am 117:3841–3852.
Jerger J. (1962) The SISI test. Int Audiol 1:246–247.
Robertson D, Irvine DR. (1989) Plasticity of frequency organization in auditory cortex of guinea pigs with partial unilateral
deafness. J Comp Neurol 282:456–471.
Kluk K, Moore BCJ. (2005) Factors affecting psychophysical tuning
curves for hearing-impaired subjects. Hear Res 200:115–131.
Robles L, Ruggero MA. (2001) Mechanics of the mammalian
cochlea. Physiol Rev 81:1305–1352.
Kluk K, Moore BCJ. (2006a) Dead regions and enhancement of
frequency discrimination: effects of audiogram slope, unilateral
versus bilateral loss, and hearing-aid use. Hear Res 222:1–15.
Schuknecht HF. (1993) Pathology of the Ear (2nd ed.). Philadelphia: Lea and Febiger.
Kluk K, Moore BCJ. (2006b) Detecting dead regions using
psychophysical tuning curves: a comparison of simultaneous
and forward masking. Int J Audiol 45:463–476.
Kohlrausch A, Fassel R, Dau T. (2000) The influence of carrier
level and frequency on modulation and beat-detection thresholds
for sinusoidal carriers. J Acoust Soc Am 108:723–734.
Levitt H. (1971) Transformed up-down methods in psychoacoustics. J Acoust Soc Am 49:467–477.
McDermott HJ, Lech M, Kornblum MS, Irvine DRF. (1998)
Loudness perception and frequency discrimination in subjects
with steeply sloping hearing loss: possible correlates of neural
plasticity. J Acoust Soc Am 104:2314–2325.
Miskolczy-Fodor F. (1960) Relation between loudness and
duration of tonal pulses. III. Response in cases of abnormal
loudness function. J Acoust Soc Am 32:486–492.
Moore BCJ. (2001) Dead regions in the cochlea: diagnosis,
perceptual consequences, and implications for the fitting of
hearing aids. Trends Amplif 5:1–34.
Moore BCJ. (2003) An Introduction to the Psychology of Hearing
(5th ed.). San Diego: Academic Press.
Moore BCJ. (2004) Dead regions in the cochlea: conceptual foundations, diagnosis and clinical applications. Ear Hear 25:98–116.
Moore BCJ. (2007) Cochlear Hearing Loss: Physiological, Psychological and Technical Issues (2nd ed.). Chichester: Wiley.
606
Sek A, Alcántara JI, Moore BCJ, et al. (2005) Development of a
fast method for determining psychophysical tuning curves.
Int J Audiol 44:408–420.
Shannon RV, Zeng F-G, Kamath V, et al. (1995) Speech
recognition with primarily temporal cues. Science 270:303–304.
Stone MA, Füllgrabe C, Moore BCJ. (2008a) Benefit of high-rate
envelope cues in vocoder processing: effect of number of channels
and spectral region. J Acoust Soc Am 124:2272–2282.
Stone MA, Moore BCJ, Greenish H. (2008b) Discrimination of
envelope statistics reveals evidence of sub-clinical hearing
damage in a noise-exposed population with ‘‘normal’’ hearing
thresholds. Int J Audiol 47:737–750.
Thai-Van H, Micheyl C, Moore BCJ, Collet L. (2003) Enhanced
frequency discrimination near the hearing loss cutoff: a consequence of central auditory plasticity induced by cochlear damage?
Brain 126:2235–2245.
Thai-Van H, Micheyl C, Norena A, et al. (2007) Enhanced
frequency discrimination in hearing-impaired individuals: a
review of perceptual correlates of central neural plasticity
induced by cochlear damage. Hear Res 233:14–22.
Thornton AR, Abbas PJ. (1980) Low-frequency hearing loss:
perception of filtered speech, psychophysical tuning curves, and
masking. J Acoust Soc Am 67:638–643.
Vickers DA, Moore BCJ, Baer T. (2001) Effects of lowpass
filtering on the intelligibility of speech in quiet for people with
and without dead regions at high frequencies. J Acoust Soc Am
110:1164–1175.