Auditory perception - Socialscientist.us

Auditory perception - Socialscientist.us
PIP_Audit-per.qxd
16/12/04
4:40 PM
Page 1
Auditory perception
Christopher J. Plack
●
The nature of sound
Basic properties.
●
From air to ear
How sound is converted into neural
activity.
●
From ear to cortex
How sound is represented and
analysed by auditory neurons.
●
Loudness and intensity
perception
Internal representation of intensity.
●
Pitch
Perception of periodicity.
●
Spatial hearing
Locating sounds in space.
●
●
Organisation of sounds
How sound is produced and
propagated
Sound pressure, intensity, and level
Pure and complex tones, frequency,
and the spectrum
Anatomy of the peripheral auditory
system
Frequency analysis in the cochlea
The role of hair cells—transduction
The auditory nerve
Rate-place coding
Phase locking
The auditory pathways: ascending
and descending
Audiograms
Explanation of loudness
Detecting changes in intensity
Pure tones: rate-place and
temporal codes
Complex tones, resolved and
unresolved harmonics
Pattern recognition theory
Temporal theory
Modern models and current research
Interaural time differences
Interaural level differences
Role of the pinna
Segregation of the sounds from
different sources.
Simultaneous and sequential grouping
Continuity illusion
Sound identification
Overview
A short summary
Text © 2004 Psychology Press Ltd. Figures © 2004 Christopher J. Plack
1
PIP_Audit-per.qxd
16/12/04
4:40 PM
Page 2
2 Psychology: an international perspective
For most of us, hearing is central in our interactions with people and with our environment.
Despite the rise of the internet, email, and text messaging, speech remains the most important form of human communication, allowing us to communicate our thoughts and feelings
with each other with great subtlety. Music is also an integral part of most of our daily lives,
whether we listen to it, play it, or are plied with it by advertisers. Furthermore, sounds of
all kinds help us interact with the environment. Some sounds warn us of danger, some wake
us up in the morning, and others provide information about the operation of machines such
as car engines and microwave ovens.
This section will begin with a description of the basic properties of sound, introducing
the slightly tricky concept of the spectrum that is essential to an understanding of the
auditory system. We will then consider how the ear analyses and then encodes acoustic
information in the form of nerve impulses that can be processed by the brain. Finally, we
will consider our perception of some of the basic acoustic properties that enable us to
segregate and identify sounds.
THE NATURE OF SOUND
Production and propagation
Sound is carried by pressure variations in some material, which can be a solid, a liquid,
or a gas. Because we are most familiar with sound as pressure variations in the air, this
section will focus on how sound waves are produced in, and transmitted through, the air.
Molecules in the air are in constant motion. Any material that is in contact with the air
experiences a pressure produced by bombardment from the molecules. When an object like
a musical instrument or the human vocal cords vibrates, this creates a disturbance in the
air. When the object is in an outward phase of vibration, the air molecules are compressed
together, and this creates a region of high pressure. In the inward phase, the air expands to
fill the vacated space, creating a region of low pressure. These pressure variations are called
sound waves.
Sound waves travel through the air around the sound source in a similar way as water
waves travel from a disturbance on the surface of a pond. When you drop a pebble into
a pond, a ring of waves travels outwards from the initial splash, expanding in two dimensions. Similarly, a disturbance in the air causes sound waves to travel outwards, although
in this case the waves radiate in three dimensions around the disturbance. Someone listening to the sound experiences a sequence of pressure variations as the sound waves pass
by, just as someone on the edge of the pond would experience a sequence of water waves,
with alternating peaks and troughs. At atmospheric pressure, sound waves travel at about
330 metres per second.
Pressure, intensity, and level
KEY TERMS
Absolute threshold: the lowest
detectable level of a sound in the
absence of any other sounds.
Decibel (dB): a unit of sound
level. The level difference
between two sounds in dB is
equal to 10 times the natural
logarithm of the ratio of the
two intensities.
The pressure variations that we usually experience are incredibly tiny. For example, a
sound near absolute threshold (the lowest sound level that we can hear) has pressure variations less than one billionth of atmospheric pressure. This is equivalent in scale to a wave
1 millimetre high on an ocean 1000 kilometres deep. The loudest sound we can listen to
without pain has pressure variations a million times greater than this, although these
variations are still small compared to atmospheric pressure. The intensity of a sound wave
(the amount of energy transmitted per second through a given area, for example, a square
metre of air) is proportional to the square of the pressure. A sound near pain threshold has
a pressure a million times greater than a sound near absolute threshold, so the auditory
system can operate over a range of intensities of a million million (a million squared)!
Since the range of sound intensities that we hear is so large, sound levels are usually
expressed in logarithmic units called decibels (dB). A constant increase in dB corresponds
to a constant multiplication of the sound intensity. An increase by 10 dB corresponds to
an increase in intensity by a factor of 10. Similarly, an increase by 20 dB corresponds to
an increase in intensity by a factor of 100, and an increase by 30 dB corresponds to an
PIP_Audit-per.qxd
16/12/04
4:40 PM
Page 3
Auditory perception
100
120
(1000)
Ozzy Osbourne
110
90
100
80
60
20
50
10
40
0
30
Sound level (dB SPL)
30
dB
Intensity ratio
90
70
Heavy traffic
80
70
Normal conversation
60
50
40
20
Quiet conversation
Library
30
1
10
0
20
Quiet forest
10
0
Absolute threshold at 1000 Hz
An illustration of how intensity ratios correspond to the dB scale (left) and an illustration of the sound pressure
levels for some familiar environments (right). Click on the loudspeaker symbols to hear sound. © Christopher J.
Plack.
increase in intensity by a factor of 1000. The factor of 1,000,000,000,000 in terms of
intensity can now be expressed by a range of 120 dB. A scale like this needs a reference
point, and 0 dB SPL (sound pressure level) is defined as the level of a sound wave with a
pressure of 0.00002 Newtons per square metre, which is roughly the lowest sound level
that we can hear.
Pure tones, frequency, and the spectrum
KEY TERMS
Pure tone: a sound with
sinusoidal variation in pressure
over time.
Waveform: the waveform of a
sound is the pattern of the
pressure variations over time.
Frequency: the number of
periods of a sound wave in a
given time measured in cycles
per second, or Hertz (Hz).
Pressure
The simplest sound wave from an acoustician’s viewpoint is called the sine wave or
pure tone. A pure tone is roughly the sound you get when you whistle a note. A pure tone
is characterised by sinusoidal variations in pressure with time (a sinusoid is defined by the
sine function in mathematics, which cycles back and forth between peaks and troughs).
If you whistle from a low note to a high note, the time between the peaks in the sound
wave decreases, and the number of peaks in pressure passing a point in space every second
increases. The time between peaks is called the period of
the wave, and the number of repetitions of the waveform
every second is called the frequency of the wave (expressed
Period
in cycles per second, or Hertz; abbreviated to Hz). The
frequency (in Hz) is equal to one divided by the period
(in seconds).
Obviously, most sound waves with which we are
familiar are much more complex than simple pure tones.
Time
However, any sound wave, no matter how complex, can be
produced by adding together pure tones with different
Wavelength
frequencies. It follows, then, that any sound wave can be
described in terms of a number of pure-tone components
with different frequencies, levels, and temporal alignments
(phases). “Bright” sounds, such as the crash of a cymbal,
contain mainly high-frequency components. “Warm”
Distance
sounds, such as a bass drum, contain mainly low-frequency
components. (As a rule of thumb, the more rapidly The sinusoidal pressure variations over time and distance for a pure tone.
a complex waveform wiggles up and down, the more © Christopher J. Plack.
3
PIP_Audit-per.qxd
16/12/04
4:40 PM
Page 4
4 Psychology: an international perspective
intense are the high-frequency components.) The spectrum of a sound can be represented by
plotting the levels of the different frequency components that are present in the sound. Light
too contains different frequency components, which we experience as different colours (red,
yellow, green, blue, etc.), and the spectrum of the light reaching us from the sun is dramatically revealed in the rainbow.
A sound wave that repeats over time, such as a vowel sound or the sound made by
a tonal musical instrument, has a discrete set of frequency components, each an integer
multiple of the repetition rate of the waveform (the number of times the waveform repeats
itself or cycles in a second, also called the fundamental frequency). These components are
called harmonics, and a periodic sound wave that has more than one component is called
a complex tone. If a complex tone has a fundamental frequency of 100 Hz, for example, the
spectrum may contain harmonics with frequencies of 100 Hz (first harmonic), 200 Hz (second harmonic), 300 Hz (third harmonic), 400 Hz (fourth harmonic), and so on. However,
it is not necessary that all these harmonics are present for the sound to have the same
fundamental frequency. For example, the fundamental frequency is unchanged (although
the pattern of the waveform that is repeating is altered) if the first harmonic is removed.
Waveforms
Spectra
1
+
0 200 400 600 8001000
+
0 200 400 600 8001000
3
Level
Pressure
2
+
0 200 400 600 8001000
4
0 200 400 600 8001000
=
1+2+3+4
KEY TERMS
Spectrum: the distribution
across frequency of the
pure-tone components that
make up a sound wave.
Fundamental frequency: the
repetition rate of the waveform
of a complex tone.
Harmonics: the pure-tone
frequency components that
make up a complex tone.
Complex tone: a tone
composed of a number of
different pure-tone frequency
components, usually with
frequencies that are integer
multiples of the fundamental
frequency.
0
5
10
15
Time (ms)
20
0 200 400 600 800 1000
Frequency (Hz)
2+3+4
0
5
10
15
Time (ms)
20
0 200 400 600 800 1000
Frequency (Hz)
A complex tone can be produced by adding together pure tones at harmonic frequencies. Notice that even if
the first harmonic is removed (lower panel) the repetition rate of the complex remains the same (200 Hz, one
period every 5 ms). Click on the loudspeaker symbols to hear sound. © Copyright Christopher J. Plack.
PIP_Audit-per.qxd
16/12/04
4:40 PM
Page 5
Auditory perception
Waveforms
Spectra
Pure tone
Pressure
Level (dB)
80
0
5
10
Time (ms)
15
Level (dB)
20
30
Time (ms)
40
Level (dB)
20
30
Time (ms)
2000
3000
Frequency (Hz)
4000
5000
1000
2000
3000
Frequency (Hz)
4000
5000
1000
2000
3000
Frequency (Hz)
4000
5000
40
20
80
Pressure
10
1000
60
0
0
50
Wide-band noise
0
20
80 Harmonic
Pressure
10
40
0
0
20
Complex tone (vowel /a/)
0
60
40
50
60
40
20
0
0
The waveforms and spectra of three sounds. Notice that the spectrum of the complex tone contains a number
of regular peaks corresponding to harmonic frequency components. Click on the loudspeaker symbols to hear
sound. © Christopher J. Plack.
A sound wave that does not repeat over time contains a continuous distribution of
frequency components. These include impulsive sounds, such as clicks, and noise, which
is characterised by a random sequence of pressure variations over time. The background
“hiss” of a radio is a good example of noise, as is the sound produced by a waterfall.
Noise can be filtered to contain a range of frequency components in any part of the
spectrum.
FROM AIR TO EAR
The peripheral auditory system
When most people talk about the “ear” they are actually referring to the pinna, the
cartilaginous flap that protrudes from the side of the head. In fact, the pinnae are not
very important to human hearing, although they do help us localise sounds to some extent
by modifying the incoming sound waves in different ways depending on the direction of
the sound source (see the next page). If someone were to cut your pinnae off, you would
not be made deaf.
Sound waves enter the ear via the ear canal, a short crooked tube that ends at the
eardrum. The ear canal has a resonance, just like an organ pipe, and this makes us most
sensitive to sound frequencies between about 1000 and 6000 Hz. The eardrum is a taut
membrane, and when sound waves reach the eardrum they cause it to vibrate. The more
intense the sound, the greater is the amplitude of the vibration. Quiet sounds produce very
5
PIP_Audit-per.qxd
16/12/04
4:40 PM
Page 6
6 Psychology: an international perspective
small movements: the eardrum moves less than the width of a hydrogen atom for sounds
near absolute threshold.
The vibrations of the eardrum are then carried through the middle ear by three
tiny bones (the smallest in the body) called the ossicles (malleus, incus, and stapes).
The stapes is connected to another membrane: the oval window of the cochlea. The
cochlea is a narrow, fluid-filled tube, coiled up into a spiral. The cochlea is where the
pressure variations are converted, or transduced, into electrical activity in neurons of
the auditory nerve. It is here that the acoustic information is translated into the language
of the brain.
Running along the length of the cochlea is the basilar membrane. When the acoustic
vibrations arrive at the cochlea they cause pressure variations in the cochlear fluids. These
pressure variations cause the basilar membrane to vibrate up and down, though in a quite
remarkable way. In an ingenious series of experiments that resulted in the award of the
1961 Nobel Prize in Physiology and Medicine, von Békésy (1947) used stroboscopic light
to view the motion of the basilar membrane in ears extracted from dead humans and
Semicircular canals
Oval window
Incus
Malleus
Pinna
Stapes
Round window
Temporal bone
Auditory nerve
Ear canal
Concha
Cochlea
Eardrum
Eustacian tube
The anatomy of the human ear. © Christopher J. Plack.
APEX
Scala vestibuli
Tectorial membrane
Scala media
Stereocilia
Scala tympani
Reissner‘s membrane
Auditory nerve
Basilar membrane
Auditory nerve
A cross-section of the cochlea. © Christopher J. Plack.
Outer hair cells
Inner hair cell
BASE
Basilar membrane
PIP_Audit-per.qxd
16/12/04
4:40 PM
Page 7
Auditory perception
animals. He sprinkled silver particles onto the membrane
to improve visibility. When he played a pure tone to
these isolated ears it caused a maximum of vibration at a
particular place on the basilar membrane, and the place
stimulated depended on the frequency of the tone. In other
words, each place on the basilar membrane is tuned, just
as each wire on a piano is tuned to play a particular note.
The basilar membrane at the base of the cochlea
(closest to the stapes) is thin and stiff and is most sensitive
to high-frequency pure-tone components. The basilar
membrane near the tip or apex of the cochlea is thick and
loose and is most sensitive to low-frequency components.
These properties vary continuously along the length of the
membrane so that each place is most sensitive to a particular characteristic frequency. A given place on the basilar
membrane will vibrate most strongly to a pure tone close
to its characteristic frequency. As the frequency of the pure
tone is moved away from the characteristic frequency
the response diminishes. Thus each place on the basilar
membrane acts as an auditory filter, selectively responding
to a narrow range of frequencies. The basilar membrane
as a whole can be regarded as a bank of auditory filters,
each tuned to a different characteristic frequency. The
range of frequencies to which each auditory filter responds
varies with frequency. Very roughly, each filter will give
a high response to a range of frequencies covering 10% of
the characteristic frequency.
Frequency selectivity
2000 Hz
BASE
APEX
200 Hz
BASE
APEX
The pattern of vibration on the basilar membrane (stretched straight for
clarity) in response to pure tones of two different frequencies. Each place
on the basilar membrane that is excited moves up and down, but not in
the same phase, so that the resulting pattern appears a little like a wave
on water (travelling from base to apex). The dotted lines show the
maximum amplitude of the vibration. © Christopher J. Plack.
2000 Hz
500 Hz
1500 Hz
BASE
375 Hz
3000 Hz
APEX
250 Hz
16000 Hz
750 Hz
1000 Hz
8000 Hz
4000 Hz
6000 Hz
What do these properties of the basilar membrane mean
for the way we hear sounds? Earlier we learned that The distribution of characteristic frequencies along the basilar membrane
any complex sound wave can be described in terms of (looking down on the cochlear spiral). © Christopher J. Plack.
a number of pure-tone components, which, when added
together, produce the complex waveform. The main function of the basilar membrane
How (and why) does the ear
break sound down into
is to separate out (to a limited extent) the different frequency components in
different frequency
a complex sound. When we hear a sound with high-frequency components, such as a cymcomponents?
bal, the basilar membrane vibrates most strongly near the base. When we hear a sound
with low-frequency components, such as a bass drum, the basilar membrane vibrates most
strongly near the apex. However, the basilar membrane has much more resolution than
this example suggests. In fact, the basilar membrane can separate out the individual
harmonics of a complex tone (for low harmonic numbers, see below).
In a sense, the ear transforms the sound waves entering the ear into a sort of auditory
KEY TERMS
rainbow, just as water droplets (or a prism) separate out the different frequency compoCharacteristic frequency: the
nents in light to give us the visual version. This is a fundamental step in the perception of
pure-tone frequency to which
sound. Sounds are identified largely by their spectra, and by the way their spectra change
a given place on the basilar
over time. Different speech sounds, such as vowels, have characteristic spectra. Because
membrane, or a given neuron
the basilar membrane represents sounds spectrally, the auditory system is able to detect
in the auditory system, is most
sensitive.
the differences in frequency composition that identify sounds. Furthermore, our ability to
Auditory filter: a term used to
separate out sounds that occur together (for example, two people speaking at once) is
refer to the tuning properties (or
largely dependent on our ability to separate out the different frequency components of
frequency selectivity) of a single
these sounds. As long as the separation in frequency is sufficient, we are able to “hear
frequency channel in the
out” one frequency component in the presence of another.
auditory system. The peripheral
As an example, consider two pure tones, one more intense than the other. When the
auditory system behaves as an
array of auditory filters, each
pure tones have a similar frequency, it is hard to hear the quieter tone
(click on the
passing a narrow range of
speaker to hear sound). The quiet tone that you are trying to hear (called the signal) is
frequency components.
masked by the more intense tone (called the masker). When the tones have very different
7
PIP_Audit-per.qxd
16/12/04
4:40 PM
Page 8
8 Psychology: an international perspective
Helmholtz: The ear as a frequency analyser
Hermann von Helmholtz (1821–1894) was one of the greatest
scientists of the nineteenth century. He made important contributions
to general physics (particularly the principle of energy conservation),
fluid dynamics, electrodynamics, optics (including the invention of the
ophthalmoscope), visual physiology, physical acoustics, and the relation
between music and auditory physiology. Previously, Joseph Fourier
(1768–1830) had proved that it is possible to produce any periodic
waveform (such as the sound produced by a vibrating string) by
adding together pure-tone components with frequencies that are
integer multiples of the repetition rate. Helmholtz knew that it was
possible to “hear out” these harmonics individually. He proposed that
the human ear behaves as a frequency analyser, that can separate out
the different pure-tone frequency components of a complex sound.
Later experiments have confirmed his theory, and have measured
more accurately the limitations of frequency selectivity in humans.
Fletcher (1940) conducted masking experiments with noise maskers and
pure-tone signals to show that only noise components close to the
signal frequency contributed to masking. He coined the term “critical
bandwidth” to refer to this range of frequency components. Following
Helmholtz, Fletcher suggested that the ear behaves like a bank of
overlapping bandpass filters (now called auditory filters), each tuned
to a particular frequency and allowing through a limited range of
frequencies. Measurements of basilar-membrane vibration in other
mammals correspond well with modern behavioural measures in
humans. The results suggest that the frequency selectivity of the human
ear is dependent on the mechanical properties of the basilar membrane.
What is psychoacoustics?
The physiological basis of hearing can be studied in other mammals,
such as the guinea pig or the chinchilla. Following strict ethical
guidelines, physiologists can record from auditory neurons and can
make measurements of the response of the basilar membrane. These
experiments are limited, however, because they do not inform about the
sensations that the animal experiences when it listens to sounds. What
we really want to know is how our (human) perceptions are related to
the physical characteristics of sounds and to the auditory mechanisms
that process these sounds. This is where psychoacoustics comes in.
Psychoacoustics is the behavioural study of hearing, in which conscious
(usually human) listeners are required to make perceptual decisions
regarding the sounds that are played to them. This may involve
discriminations between sounds that differ in some way (for example, to
find the smallest detectable change in frequency, or the lowest level
of a signal that can be heard) or it may involve subjective judgements
of the loudness or location of a sound, for example. Although
psychoacousticians usually find themselves in psychology departments,
psychoacoustic techniques can be used to study everything from the
motion of the basilar membrane, to neural coding, to more traditional
areas of psychology such as memory and attention.
It is possible that the term “psychoacoustics” was first used during
the Second World War. A secret US government project was set up to
investigate, in part, the potential of acoustic weapons. During the course
of the programme, a Dr T.W. Forbes remarked that their field could
be called “psycho-acoustics” and the name has stuck (Burris-Meyer &
Mallory, 1960). It should be noted that the principal aim of the
project was unsuccessful: the acoustic death-beam was not invented!
KEY TERM
Frequency selectivity: the
ability of the auditory system
to separate out the frequency
components of a complex sound.
frequencies, however, it is easy to hear the signal
(click
on the speaker to hear sound). This is because the place on
the basilar membrane that responds best to the signal does
not respond very well to the masker, as it has a different
frequency. It is thought that the auditory system can
selectively “listen” to the vibration at a single place on the
basilar membrane. In this way we can detect the signal and
ignore the masker.
In psychoacoustic experiments, listeners are asked
to make decisions about sounds that are presented to
them, usually over headphones in a soundproof booth.
Experiments like these can be used to measure the
frequency selectivity of the human auditory system. In one
technique, a brief (say 4 millisecond) pure-tone signal is
fixed at a low level, and played just after a longer duration
pure-tone masker. The level of the masker is adjusted until
the signal can no longer be detected. When the masker and
the signal have the same frequency, the level of the masker
needed is low. When the masker is changed to a different
frequency, the level of the masker needed is high. This is
because the place on the basilar membrane that responds
best to the signal responds less well to a masker with a
different frequency. Hence, the masker level needs to be
higher to compensate. A plot of the masker level needed
against the frequency of the masker describes a “tuning
curve”, which is a measure of the frequency selectivity
of the auditory system. Tuning curves like this can be
measured for a number of different signal frequencies,
thereby characterising the properties of different places on
the human basilar membrane.
The hair cells and transduction
Sitting on top of the basilar membrane, and arranged along
its length, are rows of hair cells. Hair cells have tiny hairs,
or stereocilia, protruding from their tops. The rows of hair
cells closest to the outside of the cochlea are called outer
hair cells. There are around 12,000 of these cells in the
human cochlea (Møller, 2000). Outer hair cells are thought
to modify the vibration of the basilar membrane, making us
more sensitive to low-level sounds and “sharpening up” the
tuning of the membrane so that we are better at separating
frequency components. Dysfunction of, or damage to, the
outer hair cells is the main cause of deafness.
The innermost row of hair cells consists of the inner
hair cells, of which there are about 3500 in the human
cochlea. These cells convert the motion of the basilar
membrane into electrical signals. As the basilar membrane
moves up and down, the stereocilia sway from side to side.
When they sway in one direction (towards the outside of
the cochlea), protein filaments connecting the stereocilia
stretch, and this opens tiny plugholes in the ion channels
in the stereocilia. Positively charged potassium ions enter the cell and cause a voltage
change in the cell. The voltage change triggers the release of a chemical neurotransmitter,
which diffuses across the tiny gap between the hair cell and a neuron in the auditory
nerve. The arrival of the neurotransmitter produces electrical impulses in the neuron. In
this way the acoustic signal is converted into an electrical signal that travels up the
PIP_Audit-per.qxd
16/12/04
4:40 PM
Page 9
Auditory perception
FROM EAR TO CORTEX
90
Masker level at threshold (dB SPL)
auditory nerve to the brain. Inner hair cells are thus the
interface between the acoustic world and the perceptual
world.
Because each inner hair cell responds to vibration at
a single place on the basilar membrane, each inner hair
cell is also tuned, responding best to a pure tone at its
characteristic frequency. Since each neuron in the auditory
nerve is connected to a single inner hair cell, each neuron
is also tuned, and carries information about the vibration
of a single place on the basilar membrane.
80
70
M
60
S
50
Freq.
40
30
MS
20
Freq.
10
0
1
The representation of frequency and
intensity in the auditory nerve
There are around 30,000 neurons in the auditory nerve,
arranged in a bundle. Those neurons in the centre of the
bundle originate in the apex of the cochlea and consequently
are tuned to low frequencies. Those neurons near the edge
of the bundle originate in the base of the cochlea and consequently are tuned to high frequencies. In this way the spatial
mapping of frequency along the cochlea is transformed into
a spatial mapping of frequency in the auditory nerve.
If the intensity of a sound is increased, the amplitude
of vibration on the basilar membrane increases, as does the
displacement of the stereocilia on the hair cells. More ion
channels are opened, more ions enter the hair cell, more
neurotransmitter is released, and the neuron attached to
the hair cell produces more impulses per second. In short,
an increase in sound intensity produces a greater rate of
firing in the neurons in the auditory nerve.
Neurons in the auditory nerve cannot fire at rates
greater than about 200 impulses per second. Most neurons
reach this rate of firing below a level of 60 dB SPL. Above
this level they can no longer change their firing rates in
response to changes in intensity, and the neurons are said
to be saturated. However, a few neurons that are a little
less sensitive can continue increasing their firing rates up to
levels as high as 100 dB SPL. Virtually the entire level
range of human hearing can be represented in terms of
neural firing rates.
Each neuron in the auditory nerve carries information
about a small portion of the spectrum of the incoming
sound. The level of these frequency components is represented in terms of the firing rate of the neuron. It follows
that the complete spectrum of the sound is carried up to
the brain in the auditory nerve, mapped on to the spatial
array of auditory neurons. If the sound has more intense
high-frequency than low-frequency components, then
neurons near the outside of the bundle will fire more
strongly than neurons near the centre of the bundle. This
is called a rate-place code, since the spectral information is
represented in terms of firing rate as a function of place in
the neural array.
9
2
3
4
5
6
Masker frequency (kHz)
A tuning curve, showing the masker level needed to mask a 4000 Hz
pure tone signal (S) as a function of masker (M) frequency. The greater the
frequency separation between the masker and the signal, the less masker
excitation on the basilar membrane at the place tuned to the signal, hence
the higher the masker level required. Data from Yasin and Plack (2003).
© Christopher J. Plack.
How is sound intensity
represented in the
auditory system?
Cochlear hearing loss and hearing aids
The hair cells in the cochlea are very sensitive and are susceptible to
damage by intense sounds and disease. Drugs such as kanamycin
(an antibiotic) and carboplatin (a chemotherapy drug) can also damage
the hair cells (Møller, 2000). In high doses, even aspirin can cause
transient hearing loss. Furthermore, as we get older there is a gradual
deterioration in hair cell function called presbycusis, which mainly
affects sensitivity to high frequencies. Cochlear hearing loss is the
most common type of deafness in the UK, affecting about one in
five people. Although dysfunction of the inner hair cells disrupts the
transduction process, making us less sensitive to the motion of
the basilar membrane, the main cause of cochlear hearing loss is
dysfunction of the outer hair cells.
Listeners with hearing loss have increased absolute thresholds
(inability to hear low-level sounds), and testing for this is the usual
means of diagnosis. However, listeners with cochlear hearing loss also
have poor frequency selectivity and an abnormally rapid growth in
loudness with level. Healthy outer hair cells are necessary for normal
frequency selectivity and for the amplification of low-level sounds that
leads to a shallow growth in basilar membrane vibration with sound
level. The abnormal growth of loudness means that although a
hearing-impaired listener may not be able to hear low-level sounds,
high-level sounds seem just as loud to them as they do to normalhearing listeners. A hearing aid that simply amplifies all sound levels
will allow the user to hear low-level sounds but will make high-level
sounds unbearably loud! Many modern hearing aids employ some sort
of compression or gain control, so that they amplify more at low levels
than at high levels (like the outer hair cells). Unfortunately, hearing
aids cannot restore the reduction in frequency selectivity. Listeners with
cochlear hearing loss find it hard to separate out the different sound
components entering their ears. This makes it hard to listen to a
person speaking in a noisy room, for example. Since there is at present
no remedy for this aspect of hearing loss, the message is simple: avoid
loud sounds and look after your ears!
PIP_Audit-per.qxd
10
16/12/04
4:40 PM
Page 10
Psychology: an international perspective
Displacement
of basilar
membrane
Scala vestibuli
Scala tympani
Time
Stereocilia
Inner hair
cell
Neurotransmitter
Auditorynerve neuron
Electrical activity in the inner hair cells and in the auditory nerve is related to the motion of the basilar
membrane. Stimulation occurs when the stereocilia are bent towards the outside of the cochlea
(to the right in this illustration). © Christopher J. Plack.
Phase locking
Voltage
Pressure
There is another way in which information about a sound is carried in the auditory nerve.
Earlier we learned that an inner hair cell is stimulated when its stereocilia are bent in one
direction, away from the centre of the cochlea. What this means is that more neurotransmitter is released, and neural impulses are more likely to be produced, when the basilar
membrane is moving in one direction. Since sound waves involve alternating high pressure and low pressure, the basilar membrane moves up and down in response. However,
nerve impulses are more likely to be produced when the basilar membrane moves just one
way. It follows that the pattern of firing is phase locked to the vibration of the basilar
membrane, and hence to the pressure variations in the air.
Take the example of the pure tone again. When
a 100 Hz pure tone is played to the ear, the basilar
membrane will vibrate up and down at 100 Hz, with the
Input sound waveform
maximum vibration near the apex of the cochlea. Neurons
connected to this part of the membrane will tend to produce impulses that are separated by 10 ms (or by integer
multiples of 10 ms), which is the period of a sound wave
with a frequency of 100 Hz. Although neurons cannot fire
at rates much greater than 200 impulses per second, they
Neural impulses
will continue to phase lock at much higher frequencies. At
these higher frequencies neurons will not fire on each and
every cycle, but the impulses will still tend to occur periodically, at the same time in the cycle of the waveform. It
is thought that this temporal synchrony occurs in response
Time
to pure tones with frequencies up to 5000 Hz. Above this
frequency, neurons cannot phase lock to the individual
Phase locking in an auditory-nerve neuron in response to a pure tone.
variations in pressure. However, they will phase lock to
Notice that although the neuron is not firing on every cycle of the tone,
peaks in the overall amplitude, or envelope, of the sound
impulses are synchronised with a particular phase of the waveform.
wave. In this case, the neurons are more likely to fire when
© Christopher J. Plack.
the sound wave has a higher overall amplitude.
PIP_Audit-per.qxd
16/12/04
4:40 PM
Page 11
Auditory perception
11
Phase locking implies that each neuron in the auditory nerve not only carries information about the gross level of activity at the place in the cochlea from which it originates,
but also carries information about the precise pattern of vibration of the basilar membrane at that place. As we will discover later, this information is used for pitch perception
and for locating sounds in space.
The ascending auditory pathway
The auditory nerve travels from the cochlea to the brainstem. The brainstem nuclei are
groups of neurons, mirrored in each hemisphere, that perform analysis on the incoming
information. The first processing station is the cochlear nucleus. The neural signal is then
passed to the superior olivary complex, which also receives input from the opposite, or
contralateral, ear. It is thought that the superior olivary complex is involved in sound
localisation. Further up the pathway are the lateral lemniscus and the inferior colliculus.
Each stage contains many different types of neurons with varying properties that decode
and analyse the auditory information. The system is massively interconnected, with
parallel pathways carrying different types of information. The information is passed to the
primary auditory cortex via the medial geniculate body of the thalamus. The primary
auditory cortex is located at the top of the temporal lobe, mostly hidden within the
Sylvian fissure.
The organisation of frequency in terms of place is carried on all the way up the
pathway to the auditory cortex. Just as different places in the visual cortex respond
to light from different points in space, different places on the auditory cortex respond to
AUDITORY
CORTEX
AUDITORY
CORTEX
MG
CN = Cochlear nucleus
SO = Superior olivary
complex
LL = Lateral lemniscus
IC = Inferior colliculus
MG = Medial geniculate
MG
IC
Cortex
IC
LL
LL
Brainstem
SO
SO
Auditory nerve
Auditory nerve
CN
CN
Cochlea
Cochlea
LEFT
Brainstem
RIGHT
The main connections in the ascending auditory pathway. The viewpoint is towards the back of the brain, as
indicated by the arrow on the illustration to the right (cerebellum removed). © Christopher J. Plack.
FRONT
BACK
LEFT
RIGHT
Sylvian
fissure
Temporal lobe
Primary
auditory
cortex
The location of the primary auditory cortex, shown from the side and in a cross-section taken along the dotted
line. © Christopher J. Plack.
KEY TERM
Phase locking: the tendency of
an auditory neuron to fire at a
particular time (or phase) during
each cycle of vibration on the
basilar membrane.
PIP_Audit-per.qxd
12
16/12/04
4:40 PM
Page 12
Psychology: an international perspective
different frequencies. However, some neurons have complex tuning properties with peaks
of sensitivity at multiple frequencies, and some have evidence of lateral inhibition, so that
certain frequencies cause the firing rate of the neuron to decrease. These neurons have
different tuning properties to neurons in the auditory nerve. It is clear that they receive
input from neurons with different characteristic frequencies, and combine the signals in
various ways. Neurons such as these may be involved in processing spectral information
across frequency and across time. Understanding exactly how this processing relates to
our perceptions is of great interest at present.
In addition to the ascending pathway, there is also a descending auditory pathway,
carrying signals from the cortex, through the brainstem nuclei, to the outer hair cells in
the cochlea. It is possible that the cortex may use this pathway, not only to influence the
processing in the brainstem, but also to influence the motion of the basilar membrane.
When the efferent neurons that innervate the outer hair cells are stimulated, the basilar
membrane becomes less sensitive to low sound levels and frequency selectivity is reduced
(Guinan, 1996). The exact purpose of this mechanism is yet to be determined, but it is an
intriguing aspect of mammalian hearing.
LOUDNESS AND INTENSITY PERCEPTION
Loudness (sones)
As we have seen, the human auditory system is very sensitive, being able to detect tiny
variations in pressure that produce tiny displacements of the eardrum. It is possible to plot
the absolute threshold of a pure tone as a function of frequency. This is called an
audiogram, and is used by audiologists to test for hearing loss. We are most sensitive to
frequencies in the range 1000–6000 Hz, and sensitivity falls off at low and high frequencies (the tone needs to be more intense for us to detect it at these extremes). The highest
frequency that can be heard by humans with excellent hearing (generally, young children)
is about 20,000 Hz, although even for these listeners the sound level needs to be very high.
As we get older we tend to lose our sensitivity to high frequencies.
Loudness refers to the subjective magnitude of a sound. The word “loudness” should
not be used to refer to a physical property of a sound. A sound may have a physical level
of 80 dB SPL, and we experience this as a particular loudness. Loudness is the perceptual
KEY TERM
correlate of sound level: as the level increases so does our experience of the magnitude of
Loudness: the perceived
the sound. However, the relation is not straightforward. Using techniques in which lismagnitude of a sound.
teners are required to compare the loudness of two sounds (for example, to adjust the level
of sound A until it seems to be twice as loud as sound B), it has been shown that loudness
scales with intensity to the power of about 0.2–0.3. Put another way, to double the
perceived magnitude of a sound, you have to increase its
level by about 10 dB, which is a factor of 10 in intensity.
100.00
For the electric guitarists among you, if you have a
100 watt amplifier and you want one that sounds twice as
10.00
loud, then you have to upgrade to a 1000 watt amplifier!
The shallow function relating loudness to intensity
implies that the huge range of intensities to which we are
1.00
exposed is compressed into a much smaller range of subjective magnitudes. This compression is mainly the result
of the outer hair cells, which effectively amplify the vibra0.10
tion of the basilar membrane in response to low-level
sounds, but not high-level sounds. This means that the
large range of sound intensities is mapped on to a much
0.01
smaller range in terms of the intensity of vibration on the
10
20
30
40
50
60
70
80
90 100
Level (dB SPL)
basilar membrane.
We have seen how the intensity of sounds is repreThe relation between loudness and sound level. Loudness is expressed in
sented in terms of the firing rates of auditory neurons, and
units called sones, where one sone is defined as the perceived magnitude
this is largely the information that gives rise to the sensaof a 1000 Hz pure tone presented at a level of 40 dB SPL. If a sound
tion of loudness. It remains a bit of a mystery, however,
appears twice as loud as this reference tone, then it has a loudness of two
sones. Based on Hellman (1976). © Christopher J. Plack.
as to why we are as good at detecting small changes in
PIP_Audit-per.qxd
16/12/04
4:40 PM
Page 13
Auditory perception
13
intensity at high levels as we are at low levels. Up to levels of at least 110 dB SPL, when
most auditory-nerve neurons are not contributing useful information because they are
saturated, we can detect changes in the level of a wideband noise (a noise with a wide
range of frequency components) of just 0.5 dB (Miller, 1947). Therefore, under experimental conditions, we can just detect the difference between a sound with a level of
100 dB SPL and a sound with a level of 100.5 dB SPL. One possibility is that we do not
use all the level information in the auditory nerve. The relatively few neurons that
contribute at high levels may be sufficient for intensity coding. The greater number of
neurons that contribute at low levels may provide more information than we actually
need, and at some stage in the auditory pathway this information may be discarded
(see Plack & Carlyon, 1995, p. 142).
It has been shown that we are good at detecting changes in the spectral shape of
sounds. For example, we can detect a “bump” in the spectrum of just a few dB. Indeed, we
can discriminate a flat spectrum from a spectrum with a small bump, even when the overall levels of the two sounds are randomised. The ability to make intensity comparisons
across frequency is called profile analysis (Green, 1988), and it is vital for identifying
sounds. Speech sounds such as vowels are characterised by the locations of spectral peaks
and dips.
Finally, the auditory system is good at following rapid changes in the level of a sound
over time. We can detect fluctuations in the level of a sound up to rates of 1000 times
a second (Bacon & Viemeister, 1985). This high temporal resolution contrasts with the
comparatively low temporal resolution of the visual system, for which the equivalent
maximum detectable rate of flicker is only around 50 cycles per second. The visual system is “sluggish” compared to the auditory system. We need the high temporal resolution
in hearing because many sounds are characterised by rapid changes in acoustic features.
This is particularly true of speech. In the visual system, on the other hand, objects are
mainly characterised by differences in their spatial properties: a high degree of temporal
resolution is not necessary.
PITCH
A sound wave that repeats over time is often heard as having a distinct pitch, which
corresponds to the repetition rate. Like loudness, pitch is a perceptual variable, and
should not be used to refer to a physical property of a sound (although it often is).
Variations in pitch are associated with musical melodies, and with the intonation contour
of the human voice. Pitch has fascinated hearing researchers for hundreds of years, but
despite all the interest, the nature of the auditory mechanisms that underlie our perceptions is still a matter for intense debate.
What is the difference
between frequency and pitch?
Pure tones
Pure tones are the simplest repetitive sound waves. For these sounds, the repetition rate is
equal to the frequency of the tone. Between about 30 Hz and about 5000 Hz, we can use
variations in the frequency of pure tones to carry musical melodies. As we have seen, there
are two ways in which the frequency of a pure tone is represented in the auditory system:
1. Rate-place code. Frequency is coded by the place on the basilar membrane (and in
other regions of the auditory pathway) that is activated.
2. Temporal code. Frequency is coded by the pattern of phase-locked firing in the
auditory nerve.
The upper limit of phase locking is around 5000 Hz, which is also around the upper limit
on our ability to perceive musical melodies (Attneave & Olson, 1971). If a familiar
melody is played using pure tones that are above 5000 Hz, the melody is not recognisable.
You can hear that something is varying, but it doesn’t sound musical. This suggests that
there may be a relation between phase locking and pitch. Furthermore, our ability to
detect a frequency difference between two pure tones is much better than would be
KEY TERM
Pitch: that aspect of sensation
whose variation is associated
with musical melodies.
PIP_Audit-per.qxd
4:40 PM
Page 14
Psychology: an international perspective
Complex tones
Complex tones also repeat over time, but the spectrum of a complex tone contains a
number of harmonics. Because the width of the auditory filters in the basilar membrane
is, roughly, a constant proportion of their characteristic frequencies, the filters become
broader as frequency increases. At high frequencies, components need to be further
Level (dB)
Spectrum of complex tone:
60
50
40
30
20
10
0
0
500
1000
1500
2000
2500
3000
3500
Frequency (Hz)
Spectral representation on basilar membrane:
Excitation (dB)
What would happen to your
hearing if your neurons lost
the ability to phase lock?
predicted on the basis of the rate-place code. We can detect a frequency difference of
just 0.2% for frequencies below 4000 Hz (Moore, 1973). The auditory filters are too
broad for this level of resolution. The frequency discrimination threshold—the smallest
detectable difference in frequency between two pure tones—increases rapidly at higher
frequencies, again suggesting that without phase locking we cannot determine puretone frequency very accurately. It appears, therefore, that the pitch of pure tones is derived
from the pattern of phase locking, although it is possible that the rate-place representation
contributes in some way, so the two systems may be working together (Loeb, White, &
Merzenich, 1983).
90
80
70
Resolved
Unresolved
60
0
500
1000
1500
2000
2500
3000
Characteristic frequency (Hz)
Basilar membrane vibration:
0
Time (ms)
14
16/12/04
10
20
30
The spectrum of a complex tone consists of a number of pure-tone frequency components (harmonics) that are
regularly spaced in frequency (top panel). The middle and lower panels show how this spectrum is represented
in the cochlea. The activity of the basilar membrane is illustrated with respect to the characteristic frequency of
each place on the membrane. In this plot, the base of the membrane is to the right and the apex to the left.
Low-harmonic numbers are separated out by the basilar membrane and there are distinct peaks of activity.
A place on the basilar membrane tuned to a low harmonic shows a sinusoidal vibration corresponding to the
waveform of the harmonic (lower panel, left). The higher harmonics are not separated into individual pure tone
components. Several harmonics interact at each place on the membrane. The result is a complex pattern of
vibration (lower panel, right) that has the same repetition rate or fundamental frequency as the original
complex tone. © Christopher J. Plack.
PIP_Audit-per.qxd
16/12/04
4:40 PM
Page 15
Auditory perception
apart in frequency to be separated by the membrane. It follows that only about the first
10 harmonics in a complex tone are separated out or resolved by the cochlea, and excite
distinct places on the basilar membrane. A place on the membrane tuned to a low
harmonic shows a pattern of vibration corresponding to the waveform of the pure-tone
harmonic.
In contrast, the higher harmonics are unresolved, and several harmonics interact at
each place on the membrane. Because of this, the vibration of the membrane is more
complex. Indeed, the pattern of vibration of each place is itself a complex tone. Of course,
the frequency spacing of the harmonics that interact at each place is the same as the
harmonic spacing for the complex tone as a whole. It follows that the repetition rate or
fundamental frequency of the vibration of the membrane is the same as that of the original complex tone.
Classical theories suggested that the pitch of a complex tone may be derived in one of
two ways:
1. Pattern recognition theory (Goldstein, 1973; Terhardt, 1974). The resolved harmonics form a pattern that is characteristic of any fundamental frequency. If harmonics of
300, 400, and 500 Hz are present, the auditory system can deduce that the fundamental frequency is 100 Hz. This mechanism requires that the harmonics are resolved,
so that their frequencies can be independently determined.
2. Temporal theory (Schouten, 1940, 1970). Pitch may be derived directly from the
repetition rate of the waveform produced by the interacting unresolved harmonics.
It turns out that a pitch can be heard when only resolved harmonics, or only unresolved harmonics are present. This means that neither of the classical theories can be a
sufficient explanation. However, the lower resolved harmonics tend to dominate pitch
perception. Complex tones containing just unresolved harmonics produce a weak (albeit
musical) pitch. A slight change in the frequency of a resolved harmonic has a much larger
effect on pitch than a change in the frequency of an unresolved harmonic (Moore,
Glasberg, & Peters, 1985). Furthermore, we can be cued to hear a pitch corresponding to
the fundamental frequency when just two successive resolved harmonics are presented to
opposite ears (Houtsma & Goldstein, 1972). Each cochlea only receives one harmonic;
therefore interaction between several harmonics on the basilar membrane is clearly not
necessary for the low pitch to be heard.
Some modern models suggest that the pitch of complex tones is derived by combining the information from the resolved and unresolved harmonics (Meddis & Hewitt,
1991; Moore, 2003, p. 223). These models can produce a good description of pitch
phenomena by assuming that the phase-locking information is analysed across characteristic frequency. Take, for example, a neuron tuned to the third harmonic of a pure tone
with a fundamental frequency of 100 Hz. This neuron will phase lock to the harmonic
frequency of 300 Hz. If it produces an impulse on every cycle, the impulses will be
separated by 3.33 milliseconds (1/300 seconds). However, the interval between every
three impulses will be 10 milliseconds, which is the period of the fundamental frequency.
In fact, for any resolved harmonic there will be intervals between impulses that correspond to the fundamental frequency. A neuron with a higher characteristic frequency,
tuned to unresolved harmonics, will phase lock to the envelope of the waveform
produced by the interacting harmonics. The intervals between impulses will also tend to
be 10 milliseconds. If the auditory system simply searches across characteristic frequency
for the most common interval between impulses, then this is likely to correspond to the
period of the complex.
It must be said, however, that the issue is far from settled, and pitch remains an
exciting field of research. The pattern recognition models have experienced something
of a resurgence recently (Bernstein & Oxenham, 2003), and there has been a suggestion
that there may be separate pitch mechanisms for the resolved and unresolved harmonics (Carlyon & Shackleton, 1994). If true, then this will seriously complicate our
endeavours!
In parallel with the psychophysical investigations, auditory physiologists have been
searching for the neural mechanisms that may be responsible for extracting pitch from
KEY TERM
Resolved harmonic: a
low-numbered harmonic of
a complex tone that can be
separated out by the auditory
system.
15
PIP_Audit-per.qxd
16
16/12/04
4:40 PM
Page 16
Psychology: an international perspective
periodic waveforms. Much of this research has focused on how the temporal pattern of
phase locking is analysed in the brain. The maximum frequency to which a neuron will
phase lock is much reduced in the medial geniculate body of the thalamus; neurons will
not synchronise to tones with frequencies above about 800 Hz (Møller, 2000).
Consequently, physiologists are looking for neurons in the brainstem that are responsible
for decoding the phase-locking information into something like a rate-place code. The
hope is to find neurons that produce their highest firing rates when they are presented
with a characteristic fundamental frequency. There is some evidence that there are neurons like this in the inferior colliculus (Langner & Schreiner, 1988).
SPATIAL HEARING
L
R
Time 1
L
R
Time 2
A sound from the right arrives at the right ear first and is louder in the right
ear. These are the main cues for sound localisation. © Christopher J. Plack.
Why do we need two ears?
The visual system has remarkable spatial resolution, with
millions of receptors in each eye responding to light
from different locations in the visual field. In contrast,
the auditory system focuses on frequency resolution, with
thousands of receptors tuned to different frequencies.
However, having two ears (effectively, two directional
receptors) helps us to locate the direction of a sound
source to some extent. There are two main cues
available:
1. Interaural time differences. A sound to the right arrives
at the right ear before the left ear.
2. Interaural level differences. A sound to the right is
more intense in the right ear.
Interaural time differences appear to be dominant for sound localisation. We can
detect an arrival time difference between the two ears of just 10 millionths of a second
(Klump & Eady, 1956)! This corresponds to a direction difference of about 1 relative to
straight ahead. If the tone frequency is a little above 750 Hz, the distance between waveform peaks (the wavelength) is less than twice the distance between the ears. This means
that for a continuous pure tone directly to the right, a waveform peak in the left ear will
(a) Pure tone
LEFT
RIGHT
L
R
(b) Modulated pure tone
LEFT
RIGHT
KEY TERM
Interaural time difference:
a difference between the arrival
times of a sound wave at the
two ears.
Time
Even though the pure tone comes from the right, it appears as if the waveform peaks lead in the left ear. If the
tone is modulated in amplitude, the ambiguity is resolved. © Christopher J. Plack.
PIP_Audit-per.qxd
16/12/04
4:40 PM
Page 17
Auditory perception
17
be closely followed by a peak in the right ear. In this way it may appear that the sound is
to the left when it is really to the right. This ambiguity can be resolved if the waveform
envelope fluctuates, so that the more slowly varying features can be compared. In general,
however, interaural time differences are more important at low frequencies.
Interaural time differences are probably extracted in the medial superior olivary
complex. Neurons have been located there that respond to input from both ears, and have
a higher rate of firing when there is a particular time delay between the two inputs
(Carr & Konishi, 1990; McAlpine & Grothe, 2003). The astonishing resolution of
the system implies that the timing of waveform features is very precisely encoded in the
phase-locked response of the auditory nerve.
In contrast, interaural level differences are more useful at high frequencies. The head
will tend to prevent the sound waves from reaching the ear furthest from the sound,
creating a sort of “acoustic shadow”, and leading to large level differences between the
ears. At low frequencies, however, the sound can bend round or diffract around the head,
so that the level differences are minimal.
Although the time and level cues provide important information about the location of
a sound, that information would remain ambiguous were it not for the help of the pinna.
For example, a sound directly in front produces the same interaural time and level differences as a sound directly behind, or indeed a sound directly above. The pinna modifies the
sound as it enters the ear, and produces peaks and dips in the spectrum that depend on
the direction of the sound (Blauert, 1997, p. 67). The auditory system can recognise these
“directional signatures” to help localisation. We also use head movements to help resolve
ambiguities. If a sound is directly in front and the head is turned to the right, then the
sound will arrive at the left ear first. If the sound is directly behind, then the same head
movement will cause the sound to arrive at the right ear first.
ORGANISATION OF SOUNDS
It is quite rare that the totality of the sound we hear comes from just one sound source.
In most cases there are several sound sources, and the sound waves from these sources are
How does the ear separate out
simply summed together in the air to produce a highly complex waveform. The auditory
sounds that occur together?
system has the difficult task of separating the sounds emanating from the source of
interest (for example, the speech from the person you are
talking to) from the sounds coming from the other sources
“C a n
I
g e t
you a
dr i
n
k?”
in the environment (for example, the other people at the
party). Bregman (1990) has termed the whole process
auditory scene analysis.
There are two aspects to this ability. First, the auditory
system must parse the sound components from different
sources that occur simultaneously. Second, it must be able
to follow the sequence of sounds from the source of interest over time. The frequency selectivity of the cochlea is
“T h is is
a
r e a lly good
p a r
t y”
crucial for the first aspect. The auditory system separates
out the different frequency components in the cochlea, The waveforms of two sentences spoken at the same time. To follow one
and then extracts and groups together those components of these sentences, the auditory system has to separate out the relevant
that come from the source of interest. To decide which speech sounds when they overlap with the competing speech, and group
components come from which source, the auditory system together the sequence of sounds over time. © Christopher J. Plack.
uses two main rules of thumb:
1. Onset times. Sound components that start together come from the same source.
2. Harmonicity. Sound components that are harmonically related (i.e., from a single
harmonic series) come from the same source. It is much easier to separate two voices
if the fundamental frequencies are different (Culling & Darwin, 1993).
To follow a sequence of sounds from the same source over time, called sequential streaming, the auditory system looks for similarity between the sounds. If the sounds are similar
in terms of their spectra, their fundamental frequencies, and their locations, then they tend
KEY TERM
Auditory scene analysis: the
perceptual organisation of
sounds according to the sound
sources that are producing them.
PIP_Audit-per.qxd
18
16/12/04
4:40 PM
Page 18
Psychology: an international perspective
to get grouped together perceptually. In this way we can attend to the sequence of sounds
from a single sound source over time. Another aspect of this process gives rise to an
auditory “illusion” called the continuity illusion. If a sound was interrupted briefly by a
second more intense sound (for example, speech interrupted by the sound of a door slamming), then it appears to us as if the first sound was continuous during the interruption.
Similarly, if a brief gap in a sound is filled with a more intense sound, then the first sound
seems continuous even though it is actually discontinuous. The sensation we experience is
an interpretation, or “best guess”, made by the auditory system on the basis of the physical evidence. When there are several sound sources competing with each other, much of
what we “think” we hear is actually a reconstruction by the auditory system.
Once the sounds of interest have been separated from the acoustic background, they
can then be identified. There are many different acoustic features to which the auditory
system is sensitive, and these apply to different types of sound. Fundamental frequency
is clearly very important for music, as is timbre, the sensation relating to the spectral and
temporal features that help characterise different instruments. For speech sounds, the
spectrum, and the way the spectrum changes over time, provides the main source of information. By focusing on a spectral analysis of sounds, the auditory system can decipher the
complex acoustic code that is the basis for human communication.
OVERVIEW
The defining characteristic of mammalian hearing is the spectral analysis performed by the
cochlea, which separates out the different frequency components of sounds. This analysis
is the basis for much subsequent processing, including pitch perception for complex tones,
and sound segregation and identification. The auditory system also encodes the temporal
patterns of vibration on the basilar membrane in the form of phase-locked firing patterns
in the auditory nerve. The precision of the temporal code gives us accurate pitch information, and enables us to localise sounds on the basis of interaural time differences. The
detailed spectro-temporal analysis, and the ability to follow rapid changes in the characteristics of sounds, provides us with a highly efficient and sensitive sensory system.
FURTHER
●
●
●
READING
Moore, B.C.J. (2003). An introduction to the psychology of hearing (5th ed.).
London: Academic Press. A comprehensive yet readable account of auditory
perception, including the basic physiology of the ear.
Yost, W.A. (2000). Fundamentals of hearing: An introduction (4th ed.). NewYork:
Academic Press. Good detail on the physics of sound, as well as anatomy,
physiology, psychophysics, and hearing loss.
Bregman, A.S. (1990). Auditory scene analysis: The perceptual organisation of
sound. Cambridge, MA: MIT Press. The definitive work on perceptual organisation
in hearing.
REFERENCES
Attneave, F., & Olson, R.K. (1971). Pitch as a medium: A new approach to psychophysical
scaling. American Journal of Psychology, 84, 147–166.
Bacon, S.P., & Viemeister, N.F. (1985). Temporal modulation transfer functions in normal-hearing
and hearing-impaired subjects. Audiology, 24, 117–134.
Bernstein, J.G., & Oxenham, A.J. (2003). Pitch salience: Harmonic number or
harmonic resolvability? Journal of the Acoustical Society of America,
113, 3323–3324.
Blauert, J. (1997). Spatial hearing: The psychophysics of human sound localization. Cambridge,
MA: MIT Press.
PIP_Audit-per.qxd
16/12/04
4:40 PM
Page 19
Auditory perception
Bregman, A.S. (1990). Auditory scene analysis: The perceptual organisation of sound.
Cambridge, MA: Bradford Books/MIT Press.
Burris-Meyer, H., & Mallory, V. (1960). Psycho-acoustics, applied and misapplied. Journal of the
Acoustical Society of America, 32, 1568–1574.
Carlyon, R.P., & Shackleton, T.M. (1994). Comparing the fundamental frequencies of resolved
and unresolved harmonics: Evidence for two pitch mechanisms? Journal of the Acoustical
Society of America, 95, 3541–3554.
Carr, C.E., & Konishi, M. (1990). A circuit for detection of interaural time differences in the
brain stem of the barn owl. Journal of Neuroscience, 10, 3227–3246.
Culling, J.F., & Darwin, C.J. (1993). Perceptual separation of simultaneous vowels: Within and
across-formant grouping by F0. Journal of the Acoustical Society of America, 93, 3454–3467.
Fletcher, H. (1940). Auditory patterns. Reviews of Modern Physics, 12, 47–65.
Goldstein, J.L. (1973). An optimum processor theory for the central formation of the pitch of
complex tones. Journal of the Acoustical Society of America, 54, 1496–1516.
Green, D.M. (1988). Profile analysis. Oxford, UK: Oxford University Press.
Guinan, J.J. (1996). Physiology of olivocochlear efferents. In P. Dallos, A.N. Popper, & R.R. Fay
(Eds.), The cochlea (pp. 435–502). New York: Springer.
Hellman, R.P. (1976). Growth of loudness at 1000 and 3000 Hz. Journal of the Acoustical
Society of America, 60, 672–679.
Houtsma, A.J.M., & Goldstein, J.L. (1972). The central origin of the pitch of pure tones:
Evidence from musical interval recognition. Journal of the Acoustical Society
of America, 51, 520–529.
Klump, R.G., & Eady, H.R. (1956). Some measurements of interaural time difference thresholds.
Journal of the Acoustical Society of America, 28, 859–860.
Langner, G., & Schreiner, C.E. (1988). Periodicity coding in the inferior colliculus of the cat: I.
Neuronal mechanisms. Journal of Neurophysiology, 60, 1799–1822.
Loeb, G.E., White, M.W., & Merzenich, M.M. (1983). Spatial cross correlation:
A proposed mechanism for acoustic pitch perception. Biological Cybernetics, 47, 149–163.
McAlpine, D., & Grothe, B. (2003). Sound localization and delay lines—do mammals fit the
model? Trends in Neurosciences, 26, 347–350.
Meddis, R., & Hewitt, M. (1991). Virtual pitch and phase sensitivity of a computer model of the
auditory periphery: I. Pitch identification. Journal of the Acoustical Society of America, 89,
2866–2882.
Miller, G.A. (1947). Sensitivity to changes in the intensity of white noise and its relation to
masking and loudness. Journal of the Acoustical Society of America, 19, 609–619.
Moore, B.C.J. (1973). Frequency difference limens for short-duration tones. Journal of the
Acoustical Society of America, 54, 610–619.
Moore, B.C.J. (2003). An introduction to the psychology of hearing (5th ed.). London: Academic
Press.
Moore, B.C.J., Glasberg, B.R., & Peters, R.W. (1985). Relative dominance of individual partials
in determining the pitch of complex tones. Journal of the Acoustical Society of America, 77,
1853–1860.
Møller, A.R. (2000). Hearing: Its physiology and pathophysiology. New York: Academic Press.
Plack, C.J., & Carlyon, R.P. (1995). Loudness perception and intensity coding. In B.C.J. Moore
(Ed.), Handbook of perception and cognition: Vol. 6. Hearing (pp. 123–160). Orlando, FL:
Academic Press.
Schouten, J.F. (1940). The residue and the mechanism of hearing. Proceedings of the Koninklijke
Akademie van Wetenschappen, 43, 991–999.
Schouten, J.F. (1970). The residue revisited. In R. Plomp & G.F. Smoorenburg (Eds.), Frequency
analysis and periodicity detection in hearing (pp. 41–54). Leiden, The Netherlands: Sijthoff.
Terhardt, E. (1974). Pitch, consonance, and harmony. Journal of the Acoustical Society of
America, 55, 1061–1069.
von Békésy, G. (1947). The variations of phase along the basilar membrane with sinusoidal
vibrations. Journal of the Acoustical Society of America, 19, 452–460.
Yasin, I., & Plack, C.J. (2003). The effects of a high-frequency suppressor on tuning curves and
derived basilar membrane response functions. Journal of the Acoustical Society of America,
114, 322–332.
19
PIP_Audit-per.qxd
16/12/04
4:40 PM
Page 20
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement