mcadams bregman streams
TRANSCRIPT
-
7/27/2019 McAdams Bregman Streams
1/21
Hearing Musical StreamsAuthor(s): Stephen McAdams and Albert BregmanSource: Computer Music Journal, Vol. 3, No. 4 (Dec., 1979), pp. 26-43+60Published by: The MIT PressStable URL: http://www.jstor.org/stable/4617866.
Accessed: 27/09/2011 13:53
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at.http://www.jstor.org/page/info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new form
of scholarship. For more information about JSTOR, please contact [email protected].
The MIT Pressis collaborating with JSTOR to digitize, preserve and extend access to Computer Music
Journal.
http://www.jstor.org/action/showPublisher?publisherCode=mitpresshttp://www.jstor.org/stable/4617866?origin=JSTOR-pdfhttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/page/info/about/policies/terms.jsphttp://www.jstor.org/stable/4617866?origin=JSTOR-pdfhttp://www.jstor.org/action/showPublisher?publisherCode=mitpress -
7/27/2019 McAdams Bregman Streams
2/21
Hearing
Musica l
Streams
Stephen
McAdams
Hearing
and
Speech
Sciences
Stanford
University
School of Medicine
Stanford,
California
94305
Albert
Bregman
Department
f
Psychology
McGill
University
Montreal,
Quebec,
Canada
Introduction
The
perceptual
ffectsof
a
soundare
dependent
pon
the musical ontext nwhich hatsounds imbedded. hat s,
a
given
ound's
erceived
itch,
imbre,
nd oudness re
n-
fluenced
by
the sounds hat
precede
t,
coincidewith
it,
and
even ollow
t in
time.
Thus,
hiscontext nfluences
he
way
a listener
will associate
he soundwith
various
melodic,
rhythmic,
ynamic,
armonic,
nd
timbral tructures ithin
the
musical
equence.
t
thus behooves he
composer
nd
interpreter
o
understand
he various
erceptual rganizing
principles
hataffect he
derivation
f musical ontext
rom
sequences
f
acoustic
vents.
We
nclude
he
interpreter
ere
because
everal
musical
dimensions,
uch
as
timbre,
attack
and
decay
transients,
nd
tempo,
are
often not
specified
exactlyby
the
composer
nd
are
controlled
y
the
performer.
In this articlewe shalldiscussprincipleshatdescribe
how
variousmusical
imensions
ffect he
perceived
ontinui-
ty
of music.
Leon
van
Noorden
as
stated hat "in
sequences
where
he tones follow
one another
n
quick
succession,
effectsare
observed
which ndicate
hat
the tones arenot
processed
ndividually
y
the
perception
ystem.
On
the one
hand
we
find various
ypes
of
mutual
nteraction etween
successive
ones,
such as
forward
nd
backward
masking,
loudness
nteractions
ndduration
nteractions. n
he other
hand,
a kind of connection
s
found
between
he
successive
perceived
ones."
[28,
p.1
].
As
for
simultaneousonic
events,
Bregman
5,
9]
has
suggested
hat
different
ounds
are
ex-
tracted
ccording
o
various
erceptual
nd
cognitive
rga-
nizational
mechanismsrom
he
superimposed
coustic ibra-
tions.While omeresearchers ould ike to attribute hese
phenomena
o
mechanisms
n the
peripheral
r
early
entral
nervous
ystem which
are
surely
nvolved o
some
extent,
see
[24,
33]),
prominent
members
f thismore
psychophys-
ically
oriented school"
re
beginning
o
think he
organiza-
tion
s
too
complex
o
be so
easily
explained.
However,
ather
than
delve
nto
theoretical
xplanations
f these
phenomena,
it will suffice
or
the
present
purpose
o describe
hem
n
general
nd o
briefly
quantify
ome
salient
parameters
hat
have
ntrigued
s
with
compositional
ossibilities.
his
artic
thuspresents,na tutorialashion, review f researchin-
cluding
ur
own)
whichhas direct
mplications
or musi-
cians,
specially
or
composers
orking
with
computer
musi
Inall of our
researchnd n
mostother
research
ited,
com-
puters
predominantly
DP-11
minicomputers)
ereused
o
synthesize
he
sounds
presented.
They
were
alsousedfor
presentation
f sound
timuli,
ollection
f
responses
rom h
listeners,
nd
analysis
f
the data.For
thorough
ummari
and heoretical
reatmentsf thisarea
of
research,
ee
[2,
5,
9.]
.
What
s An
Auditory
tream?
Auditory treamormationheory s concernedwith
how the
auditory
ystem
determines
hether
sequence
f
acoustic
vents
esults rom
one,
or
more
han
one,
"source
A
physical
source"
may
be
considered
s
some
sequence
f
acoustic vents
emanating
romone
location.
A
"stream"
is a
psychological
rganization
hat
mentally
epresents
uch
a
sequence
nd
displays
certain
nternal
onsistency,
r con
tinuity,
that allows
that
sequence
o be
interpreted
s a
"whole."
By way
of
example,
wo
possible
perceptual
r-
ganizations
f
a
repeating
ix-tone
sequence
re llustrated
n
Figure
1. Time
s
represented
n
the horizontal
xis and
frequency
s
represented
n
the
vertical
xis.
Thedotted
ine
connecting
he tones
n the
figure
ndicate
he stream
erce
In the first
configuration,
ix
tones
are
heardone
after
the
other n a continuouslyepeatingycle (Taped llustratio
la)l;
it is
easy
to
follow
the entire
melodic
pattern.
n
the
second
percept,
hough,
ne
might
hear wo
separate
hree-
tone
patterns
which
appear
o
have ittle
relationship
o
eachother
Taped
llustration
1b).
It is
difficult
n thiscase
to
follow
he
original
ix-tone
pattern.
Note
that
n
the
first
example
ne stream
s
heard,
nd n
the
second,
wo
are
hea
?
1979
by
Stephen
McAdams
nd
Albert
Bregman
(1)
A
tape
containing
ound
examples
s
available
rom
Mr.
McAdams.
escriptions
f
the
taped
llustration
are
ound
n
Appendix
.
Page
26
Computer
Music
Journal,
Box
E,
Menlo
Park,
CA
94025
Volume
3
Numb
-
7/27/2019 McAdams Bregman Streams
3/21
D
D
D
B"B
/B
B'
F
F
%
F
-
"
I
I
.
O
ST
I
I
Cr
I
I
-
%
%
(a)
(b)
Time
-,
Time
One
Stream
Two Streams
Figure
1. A repeating6-tone sequence
composed
of
interspersed
high
and low tones can resultin differentpercepts.In Figure
l
with
high
and
low tones
alternating
at a
tempo
of 5
tones/sec.,
one
perceptual
tream
is
heard.
In
Figure
1
b,
at
a
tempo
of
10
tones/sec.,
the
high
tones
perceptually
egregate
rom the low tones
to form two streams
(cf.
Taped
Illustration
1).
D
4
F
t
t
D
o
0
L.
F
C
C
u
"=I
-U
A
,/
A
/
Time
-+
Time
-+
I : J J
J i
I : J J J :
II
U
High
Stream Low
Stream
High
Stream
Low Stream
Figure
2.
Due
to the
competition
among
stream
organizations,
one
F
may
be
perceived
as
belonging
o
either the
higher
stream
or
the
lower
stream
but
not to
both.
The
organization
of
the
streams
changes,
among
other
things,
the
perceived
rhythmic
structure,
as
indicatedunder
each
diagram
cf.
Taped
Illustration
2).
Stephen
McAdams
nd Albert
Bregman:
Hearing
Musical
Streams
Pag
-
7/27/2019 McAdams Bregman Streams
4/21
In
the rest
of
this section we
will
discuss
some
of the
properties
exhibited
by
a
stream.
For
example,
it is
possible
to focus
one's
attention
on a stream and follow
it
through
time
[6,
28].
In the first
taped
example
one
can
follow the
six-tone
pattern
without
any
trouble.
Thus
a
stream
must
exhibit
a certaincoherence
over
time.
However,
n
the second
example,
it
is
difficult
to follow
the six-tone
pattern,
but
it is
easy
to
follow
either three-tone
pattern.
Notice that one
can
pay
attention
to either the
higher
or lower
stream,
switching
between
them
at
will,
but that it
is not
possible
to attend
to
both
simultaneously (Taped
Illustration
ib).
Indeed,
each
three-tone pattern n Figurelb constitutesa separate tream
and
maintains
ts own
temporal
coherence.
While
a listener
s
paying
attention
to
one
coherent
stream,
other acoustic
in-
formation
is
perceptually
relegated
to
the
background.
If
one
group
of
sounds is distinct
enough,
the
foreground
back-
ground
relation
may
be almost
involuntary
and it
may
require
a
great
deal
of attentional
effort to focus
on
streams
nitially
relegated
o the
background.
The
information-processing
ature
of the
stream
segre-
gation
process
s
suggested
by
the
observation
hat the
segre-
gation
of
a
sequence
nto
smaller treams
akes time
to
occur
[4,
13].
In
Figure
lb,
notice that
one
can
hear a
six-tone
pat-
tern for
the first
few
cycles
before
it
segregates
nto
two
separate treams TapedIllustration b). It thus appears hat
the
perceptual
system
assumes
things
are
coming
from
one
source
until it
acquires
enough
information
to
suggest
an
alternate
nterpretation.
When
temporal
coherence
is
lost in
a
sequence,
it
be-
comes
more difficult to order the
events of
that
sequence
n
time
[6,
11,
14,
28]
.
In
Figure
la,
it
would
be
easy
to tell the
orderof
tones
A,
B,
andC. But as this
larger
treambreaks
down into
the smaller
treamsof
Figure
b,
i.e.,
as
temporal
coherence
s
lost,
it
becomes
more difficult
to
judge
the order
of these tones. In such a case
one
might
notice
that tone
A
comes
before tone
C but
it
would be hard
to tell
whether
tone B
came before
A,
between
A
and
C,
or
after
C. This
statement
should be
qualifiedby noting
that in
the
tone se-
quences
mentioned
all
of the
tones are
of
equal
loudness
and
timbre.
The
tone
sequences
are furthermore
continually
recycling
and
are
faded
in
so that the
listener cannot label tone
A as
being
the first
tone in the
sequence.
Information n a
musical
context,
such
as
the
first
beat
being emphasized
as
a
downbeat,
may give
the
listener an
anchor
point
against
which
to relate
the
temporal
positions
of other
tones. Thus
one
can
judge
the order
of
events in
time
within
a
given
perceptual
streambut
not
necessarily
across
streams.
Accompanying
his
is the observation
hat different streamscan
appear
o
overlap
in
time even when
they
do
not.
Again
in
Figure
lb,
notice
that one can hear two three-tone
patterns apparently
going
on at
the same time even
though
the tones are
alternating
(TapedIllustration b).
If an
event
is
potentially
a memberof
more
than
one
"competing"
tream,
one
may perceive
t as
belonging
o one
stream
or another
but
not to both
simultaneously[3,
10,
11] .
This
is not
to
say
that
a musiciancannot
hear
several
simul-
taneous
lines. The
point
is that it is
impossible
o
use several
parsing
schemes at the
same time.
Figure
2 illustrates
the
effect
on
our second
example
of
moving
the
higher
triplet
into closer
frequency
proximity
to
the
lower
triplet.
We can
find an
intermediate
position
where
the lowest
tone,
tone
F,
can
group
with
either
the
higher
or lower
stream
(Taped
Illustration
2).
Notice
that
this
regrouping
esults n
a
rhyth-
mic transformation
as
shown under
each
configuration.
Some
non-musical
examples
of
this
phenomenon
are the
fac
vase illusion
and
the reversible
Necker
cubes;
Escher
and
Vasarely
have
produced
art works
based
on such
principle
of
perceptual
organization.
Finally,
this
brings
up
the
relationship
between
"sourc
and
"stream."
A stream
s
perceived
as
emanating
rom
a
sin
source.
So,
in
the
first
example,
the
pattern
was
fairly
con
tinuous
and was
easily
recognized
as
coming
from
one
sourc
but in Figure lb, the large frequencydistancebetween the
two
three-note
groups
introduced
a sort
of
discontinuity
that caused
the
perceptual
system
to
interpret
the
sequenc
as
resulting
rom
two sources.
Since
at
any given
moment
th
composite
pressure
variations
timulating
he ear
result from
several
ources,
the
auditory system
needs
a
battery
of heuri
tics to
parse,
or
segregate,
he
information
into
separate
streams.
It thus needs
to
build a
description
of the
acoustic
environment rom
separate
descriptions
of
the
various
stream
and
the
relationships
between
them
[2].
Factors
which the
perceptual
ystem
uses to
build
descriptions
of
streams,
and
subsequently
sources,
are
frequency,
rate
of occurrence
of
events
(or
tempo),
intensity,
timbre,
and
attack/decay
tran
sients.
In
the
rest
of the
paper
hese will be discussed
n
som
detail.
Of
course
it is obvious
that sounds
are
assigned
perce
tually
to different
sources
when
the
physical
sources
are at
different
spatial
positions.
In
this
case,
intensity, spectral,
and
temporal
cues
are
all
utilized to
parse
the
sound
into
separate
sources.
However,
we
will
primarily
confine
our
discussion
to the
illusion
of
many
sources which
occurs
due
to the
organizations
within
a
single
emanation
of sound
Frequency
and
Tempo
Consider
hat
a
repetitive
cycle
of tones
spread
over a
certain
frequency
range
may
be
temporally
coherent,
or in
tegrated,
at a
particular
empo.
It is
possible
to
gradually
n-
crease the
tempo
until certain tones
group
together
into
separate
treams
on the
basis
of
frequency,
as
discussed
abov
The faster
the
tempo,
the
greater
he
degree
of breakdown
o
decomposition
into narrower
treams
until
ultimately ever
given
frequency
might
be
beating along
in its own
stream.
This
last
possibility
is
dependent
upon
a
number
of other
factors which
will be
discussed
ater.
Figure
3 illustrates
the
possible
stages
of
perceptual
decomposition
or
a
recycling
six-tone
pattern
as
one
gradu
ly
increases
the
tempo (Taped
Illustration
3).
Note
that
streams
per
se are
not tracked
beyond
a
certain
point,
but
a
texture
or
timbre s
perceived
since
the
ability
to
temporally
resolve he
individual
ones
degenerates
altogether.
Of
cours
one couldhold the tempo constantandgradually xpandthe
frequency
relationships
o achieve
a
similar
streaming
effec
[3,
6,
14,
17],
but
the
musical
consequences
would
be
vastly
different
as
one can hear
n
Taped
Illustration
4.
Dowling [1
used
simple
melodies
to illustrate
this
frequency-based
streamingprinciple.
He interleaved
wo melodies
n
the
same
frequency
range
hereby making
t
very
difficult,
without
pr
knowledge
of the
melodies,
to
separate
them
perceptually
But as
they
were
pulled
apart
n
frequency,
.e.
when
all
the
tones
of one
of
the melodies
were
transposed
upward,
each
melody
became
apparent Taped
Illustration
5).
Page
28
Computer
Music
Journal,
Box
E,
Menlo
Park,
CA
94025 Volume 3 Numb
-
7/27/2019 McAdams Bregman Streams
5/21
r
I
?
I
r
I
I
I
i I I
I
-
I
.
"
.
.2-.
,
,
=.
cc
L
L
0
0)0
cm
Q/)
%
r-
S
IE
0
oo
E
O
-------
-----,
--------
--o,)=----------
-
CC
u,
QI
----------------------?
-I
Time
Timbre/Texture
Figure
3.
This
figure
illustrates the
decomposition
of
an
acoustic
sequence
into
smallerand
smaller
perceptual
treams
as
the
frequency
separation
between the
tones or the
tempo
of
the
sequence
increases.In the
latter
case,
a
point
is
ultimately
reachedwhere
one can
no
longer perceive
individual onal
events;
a texture or
timbre is
heard
nstead
(cf.
Taped
Illustra-
tions
3
and
4).
The
requency
ange
withinwhich he
perceptual
ystem
groups
ones on the basisof
frequency
roximity
s
not
constant;he groupinganvarywith the particular attern
of
frequencies
resented.
or
example see Figure ),
two
tones,
A and
B,
are
arranged
ith
given
frequency
nd
temporal
eparations
uch
that
they
will
always
stream
together
whenno other ones are
present.
We can create
a
similar
rganization
ith
tonesX and
Y
in another
requency
range.
These wo
groups
will
each
form
a stream
s
can
be
heard n
Taped
llustration
a.
Bybringing
he
pairs
nto
the
same
requency ange,
new streams an be
formed,
uchas
A-X
and
B-Y,
on
the basisof
an
alternative
roximity
organization
3]
(Taped
llustration
b).
Thus,
he
particular
relationships
etween
requencies
n a
tonal
pattern,
ndnot
just
the
frequency
eparation
etween
adjacent
ones,
plays
a
vital
role n
the formation f
streams.
It
appears
romour
examples
hatthere s an
essentially
inverse
and
strictly
interdependent
elationship
etween
tempo
and
frequency
elationshipsmong
ndividual
ones
[26,
28]:
the
faster he
tonesfollowone
another,
he
smaller
the
frequency
eparation
t
which
hey
segregate
nto
separate
perceptual
treams.
Conversely,
he
greater
he
frequency
separation,
he slower
he
tempo
at
which
egregation
ccurs.
This llustrates
et
another
spect
of
music
which,
with the
aid
of
the
computer,
ancomeunder
he
composer's
ontrol.
Suppose
we
make
a
graph,
s in
Figure
,
which
relates
frequencyeparation
f
two
alternating
ones
on one axisto
A&B A&B
Isolated Absorbed
A
A
je.-"
x,0.1
4
S
O
O
Y
>
B
B
a)
(D
L
X
Time
Time-
Figure
4. This
figure
illustrates
he
effect
of
frequency
cont
as
opposed
to
frequency
separation
on
stream ormation.
In
the
first
part,
tones
A
and
B
form
one
perceptual
tream
wh
is unaffectedby the simultaneous treamcomposedof tone
and
Y.
When
ones
X
and
Y
are moved
into the
proximity
o
tones
A and
B,
which
remain
unchanged,
an
alternate
perceptual
nterpretation
esults
whereby
tones
A
and B ar
assigned
o
separate
treams
on
the
basis
of
a new
frequenc
context
(cf.
Taped
Illustration
).
the rate
of
alternation
f
the
tones
on
the
other
axis.
Note
thatthe horizontal xis ndicates
ncreasing
one
repetiti
time,
which
corresponds
o
decreasing
empo.
Onecandra
boundaries n
this
graph
ndicating
he
frequency-tem
regionsnwhich hetonescohere s a single tream nd ho
in
which
hey
segregate
nto
two
simultaneous
treams f
different
requencies.
here
are
two
such
boundaries.
eo
VanNoorden
28]
has
termed hese he
ission
boundary
temporal
oherence
oundary
ndnoted
hat
they
are
slig
different oreach
person.
n
Figure
the
upper
urve
s
the
temporal
oherence
oundary.
Above his
boundary
t is
impossible
o
integrate
he
two
alternating
ones into one
stream.Below
this
boundary
ies
the
temporal
oheren
region,
where t is
possible
o
integrate
vents nto
a
sing
perceptual
tream.
The lower
boundary
s the
fission
boundary.
elow his t
is
impossible
o
hearmore
hanon
stream.Note
that
the
regions
f
fission
and coherence ls
overlap, reating n ambiguousegionwhereeitherperce
may
be
heard.
As the
tempo
decreases
emporal
oherence
an b
maintained
t
greater
requency
eparations.
owever,
he
frequency
eparation equired
or
fission remains
airly
constant
with
decreasingempo.
The
emporal
oherence
fission boundaries bove
and
below
a
given
tone
are
symmetrical
ith
respect
o
pitch.
This
ndicates
hat
the
phenomena
f
temporal
oherence nd
fission
may
occur
depending
n the
absolutemusical
nterval
etween
he
ton
but
irrespective
f
the direction
f
pitchchange.
While
he
are substantial
uantitative
ifferencesn
theseboundar
Stephen
McAdams
nd
Albert
Bregman:
Hearing
Musical
treams
Pag
-
7/27/2019 McAdams Bregman Streams
6/21
Tempo
(Tones/Sec)
20 10
5
3
o
15
Al
/
Alwaysays
c/
Segregated,
O
10 5
Tone
Repetition
TiAm
biguous
=
IRegion
Always
Fission
BoundaryCoen
0 50 100 150
200
250 300
350 400
Tone Repetition Time (msec)
Figure
5.
Boundaries of
temporal
coherence
(upper
curve)
and
fission
(lower
curve)
define three
perceptual regions
in
the
stream
relationship
between two
alternating
tones each
lasting
40
msec.
This
relationship
is
a
function of both
the
tempo
of
alternation
and
the
frequency separation
between the tones
(after [28]
).
between
listeners,
the
qualitative
rends
tend
to
be similar
[28].
Note that the difference between boundaries is
increasingly
ubstantial
for
tempi
below about
10
tones/sec.
(tone
repetition
time
=
100
msec.).
This
means that
at
very
slow
tempi
of, say,
five
tones/sec.,
a
separation
of more
than
a minor 10th is
necessary
o induce
streaming.
Below
this
it
becomes
virtually
impossible
to induce
streaming.
In the
limiting
case,
if the
temporal
distance
between tones
is too
great,
they
do
not
seem to
be
connected
at all but sound
as isolated events.
A
musically
relevant
aspect
of these
boundaries hould
be mentioned.
The
region
between
the
two
boundaries
may
be
considered
o be an
ambiguous egion
since
either
a
segregated
or an
integrated
percept
may
be heard.
The
primary
determining actorin this region s attention. In otherwords,
it is
possible
to shift one's
attention back
and forth
between
the two
percepts
when the
sequence
falls
in this
region.
For
example,
one
may
focus either on a whole
stream
percept
or
on
smaller ndividualstreams.
It
appears
that the closer
the
sequence
ies
to
one of
the
boundaries,
he
easier
t
is
to
focus
on the
percept
which
is
predominant
beyond
that
boundary.
Conversely,
t is
very
difficult in this
situation to shift
one's
attention back
to
the
other
percept.
Once
the
physical
values
go beyond
either of these
boundaries,
attaining
the
comple-
mentary
percept
may
be
considered
o be
impossible.
The
role
of
attention will be
discussed n more detail later.
FrequencyTrajectories
Another
frequency-based
effect involves
frequency
trajectories.
These are
important
on two
levels. The
first
involves
trajectories
between
tones
(see
Figure 6).
Bregma
and
Dannenbring
7]
have
found
that tones
that are
con-
nected
by
glissandi
are
much
less
likely
to
segregate
under
given
conditions
of
tempo
and
frequency
separation
than
those
which
make
abrupt requency
transitions. ntermedia
situations
beget
intermediate
results.
Yet even
when
using
sinewave for frequencymodulationof a sine tone, it is pos-
sible to discern
a sort
of
streaming
ffect
of the
higher
and
_I
I
Semi-
a I
D
I
I I
I
Ramped
I
I I
r- I
rapeady-
Tas
Sta
1
a.,
a
a
Discrete
aaTime-+
I __ _ __ _
Steady- Transition
Steady-
State State
Figure
6.
These
are
the 3
types
of
frequency
transitions
between
tones used
by
Bregman
and
Dannenbring
(1973).
In
the
top
section
the tones are
completely
connected
by
a
frequency
glissando.
In the
middle section
an
interrupted
glissando
is
directed towards
the
succeeding
tone.
No
frequency glide
occurs
in the bottom
section;
the
first tone
ends on
one
frequency
and
the next tone
begins
on
another
(after
[71 ).
The unconnected
tones
segregate
more
readily
than the
others.
lowerpeaksof the modulationat certainmodulationfreque
cies
(Taped
Illustration
7).
At
lower
modulating frequenci
one can
track
the
modulation,
but higher
modulating
fre-
quencies
result
n the
effect
of
a texture or timbre. This con
tinuum
is
found
for
modulation
involving
discrete
changes
in
frequency
as
well.
The second
type
of
trajectory
might
be
called
a melod
trajectory.
The
basic rule
goes: large
jumps
and
sudden
changes
n
direction
of a
melody produce
discontinuity
in
that
melody.
In
terms
of stream
formation,
one
or
two
tone
in
a
melody
that
are
removed
rom the
melodic
continuity
o
the
rest could
be
perceived
as
coming
from
a different sourc
Page
30
Computer
Music
Journal,
Box
E,
Menlo
Park,
CA
94025
Volume 3 Num
-
7/27/2019 McAdams Bregman Streams
7/21
and
would not be
integrated
nto
the
melody
as
a
whole,
perhaps eaving
a
rhythmic
gap
in the
phrase
depending
on
the natureof the
main
sequence.
It
has been
reported,
how-
ever,
that
the
excluded tones
are
sometimes
noticed in the
background,
with their absence
having
ittle
effect
on
the
main
melody
line
[23].
Implied
polyphony
in a
solo
compound
melody
line,
as
in the suites
for
solo
instruments
by
Bach,
is a
compositional
use
of
this
principle.
Schouten
[30]
reported
that if
an
ascending
and
de-
scending
major
scale
fragment
played
with
sine
tones is
con-
tinually
repeated,
temporal
coherence
is
maintained
up
to
a
tempo of about 20 tones/sec. However, f these same tones
are
arranged
t
random,
he
maximum
tempo
at which
coher-
ence still occurs s
reducedto
about 5-10
tones/sec.
It
might
be inferred
that
the reduction
of
predictability
reduced the
pitch
boundary
within
which the
auditory
system
could
successfully
integrate
the
incoming
information. But
van
Noorden
[28]
found
that
previous
knowledge
of
the
order
of
tones
had
no
effect on
the
coherence
boundary.
It
might
be that the
small
frequency
umps
in
Schouten's
demonstra-
tion are effective
in
holding
the
stream
ogether.
Heise and
Miller
[23] investigated
our
melodic
con-
tours: a
V-shaped
contour,
an
inverted
V,
and
rising
and
fall-
ing
scale
patterns;
he
V-shaped
patterns
which
change
di-
rection
can be
thought
of
as
being
less
"predictable"
han
the scale
patterns
which move
in
only
one
direction.
Each
pattern
was
eleven tones
long
and
the
frequency
of
the
middle
tone was
variable.
They
found
that the
degree
to
which
the
variable
one
could
be
separated
n
frequency
from the
rest
of
the
pattern
before
it
segregated
nto its
own stream
was a
function
both of
the
shape
and of
the
steepness
of
the
pattern.
The
steepness
was varied
by
keeping
the
tempo
constant
and
varying
he
interval
between
successive
ones.
As
the rate
of
frequency change
for
the
entire
pattern
increases
over
time,
so does
the
amount of
frequency separation
of the
middle
tone
required
o
producesegregation.
Less
separation
of
the
middle
tone
is
required
to
produce
segregation
with
the
V-shaped patterns
than with
the
scale
patterns,
possibly
indicatingthat the perceptualsystem can follow
"predict-
able"
patterns
more
quickly
than
"unpredictable"
ones.
(However,
certain
anomalies
n
their
data,
which will not be
discussed
here,
might
suggest
other
interpretations.)
Van
Noorden
found
similarresults
for
patterns
with as
few
as
three
tones
[28]
.He
investigated
he relative
emporal
coherence of
so-called
linear
and
angular
three-tone se-
quences.
For
linear
sequences,
i.e.,
those
with two
tone
intervals
n the
same
direction,
the
temporal
coherence
bounda-
ry
occurs
at
faster
tempi
than were
found
with
angular
se-
quences
in
which the
melodic
pattern
changed
direction
at
the middle
tone. In
this latter
case
the
first
and third
tones
are more
contiguous
in
frequency
than are the first
and
second, or second andthirdtones, which facilitatesa loss of
temporal
coherence
similar o that found
by
Schouten. It
should be
noted
that while
these melodic
trajectory
effects
do
seem to
play
a role in
musical stream
ormation
beyond
the
simple
frequency
separation ffect,
they
are
certainly
con-
founded
by
other
contextual
organizations.
Both of
these
trajectory
examples
llustrate
a
principle
of
perceptual
organization
that uses
pattern
continuity,
in some form
or
another,
as a
criterion or
"source"
distinc-
tion.
Figure
7
illustrates,
however,
that
frequency proximity
may
sometimes
compete
with
trajectoryorganization;
Deutsch
found
that
simultaneous
ascending
and
descending
cale
pat-
terns
presented
to
opposite
ears
segregate
nto
upright
and
inverted
V-shaped
melodic
contours
[16].
Each
contour
is heard as if
being
presented
o one ear. The
same result
wa
found
by
Halpern
[22]
for simultaneous
ascending
and
de
scending
sine tone
glissandi,
as can be heard
n
Taped
Illust
tion
8;
in this case both
glissandi
were
presented
o both
ea
In
these two
examples
a
stream
boundary
s establishedat
the
pitch
where
the two lines
cross.
a
a
High I
\
Contour
I
t
1
--I
S
Low
.Contour
L
a
/ Low
'
_
Time-+
Figure
7. This
stimulus,
used
by
Deutsch
(1975),
is
com
posed
of
ascending
and
descending
scales
presented
sim
taneously
to
opposite
ears. Two
V-shaped
patterns
(hi
and low contours outlined
with
dashes)
are
perceived
rath
than the
complete ascending
and
descending
cale
pattern
Loudness
nd
Continuity
ffects
Fission
an be obtained
by alternating
equences
f
tonessimilar
n
frequency
nd
imbrebut
differing
n
inten
ity.
Figure
8
illustrates
he
range
of
percepts
achieved
as
the
amplitude
evelof toneA is varied elativeo thatof toneB
in the
alternating
equence
ABAB...
.
The reference
ev
for toneB used
by
vanNoorden
n these
experiments
28]
was35 db
SPL;
he
frequency
f both toneswas
1
KHz
an
each
tone
lasted40 msec.
If the levelof tone A is
below
the
auditory
hreshold
approximately
db
SPLat
1
KHz
only
the B streams heard
t half
tempo,
as
might
be
expec
(see
Figure a).
When
one A is loud
enough
o be heard
n
is at least
5
db below
tone
B,
two
separate
treams f
dif
ferent loudness can be
perceived,
each at half
tempo,
with
A
being
the softer stream
(see Figure
8b).
When
A is within
5
db of
B,
a
"pulsing"
tream
s heard and neither the
A
nor
the
B
streamcan be heard
ndependently,
.e.,
tempora
coherences inevitablen thisrangesee Figure c). As th
level
of
the A tone
s
increased bove
hatof the
B
tone,
th
different
percepts
may
result,
depending
n
the
alternati
tempo
of A andB. If this
tempo
s lessthanabout
13
tones
sec.,
fission s the next
percept
eard.This ime
B is the
sof
stream
see Figure
d).
Thusa certain
degree
of
loudness
difference llowsus to focus
on eitherstream
using
only
this
nformation.
f
the
tempo
s
greater
hanabout
12.5 1
tones/sec.,
he
percept
encountereds the "roll"
effect
discovered
y
vanNoorden. t
soundsas if stream
A,
the
louder
tream,
were
pulsing
t
half
tempo
as in the
fissio
percept,
ut the B
stream ounds
s f it were
pulsing
t
ful
Stephen
McAdams
nd Albert
Bregman:
Hearing
Musical
treams
Pag
-
7/27/2019 McAdams Bregman Streams
8/21
A
Below Threshold
a
-K
A
B
(
B
time
Fission
b
fA
B
A
I
Coherence
c
A
B
A
B
Fission
d
A
B
A
B
Roll
e
A
B
A
B
Continuity
Masking
g
A
B,
A
I
Figure
8.
This
figure
illustrates
the
range
of
possible per-
cepts
found
by
van Noorden
(1975)
for
two
alternating
ine
tones
differing
in
each case
only
in
their
intensities.
Tone
B
was
kept
at 35 db SPL
throughout;
both tones were 40
msec.
long
and
had
a
frequency
of
1
KHz.
a)
The level
of
tone
A
is
below the
auditory
threshold.
b)
Tone A is
at
least
5
db below
tone B.
c)
Tone A
is
within
5 db
of
tone B.
d)
Tone
A
is
louder
than tone B with an
alternation
empo
less
than about
13
tones/sec.
e)
Tone A
is louder
than tone
B with more
than
13
tones/sec.
f)
Tone
A
is
about
18-30
db
louder
than
tone B
and
the
tempo
is still above
13
tones
/sec.
g)
Tone
A
is
more
than 30
db louder
than
tone
B.
The
arrows
indicate
the
percepts reported;
a more
complete
description
s
given
in
the
text.
tempo
(see
Figure
8e).
Thus
the A stream
may
be heard
n-
dependently,
but
not
the
B
stream.
In other
words,
t is as
if
the A tones consisted
of
two
parts:
one
that combines
with
the B
stream
to
give
a
full
tempo
roll,
and
another,
at half
tempo,
which can
be
perceived
separately.
At a
tempo
of
about
13
tones/sec.
another effect
emerges
when
the level
of A
is about
18-30 db above that
of
B.
This is the
continui
effect
shown
in
Figure
8f,
so
namedbecausetone
B
is
not
heardas
pulsing
but ratheras a
fairly
soft,
continuous
tone
under the
louder,
pulsing
A
stream;
this is an
example
of a
class of effects
that will be
discussed below.
Finally,
if the
levelof the A stream s incremented till further, his stream
completely
masks
the B
stream
Figure 8g).
Again,
this
set o
loudness-based
phenomena
exhibits an
ambiguous region
between the coherence and fission boundaries
where one
might pay
attention to either the A or B
streams
ndividuall
or
to the AB
stream as
a
whole.
There are
thus
three
per-
ceptual
regions
for
alternating
ones
at
the same
frequency
where the
tempo
is above about 12.5
tone/sec.:
the
roll
regi
the
continuity region,
and the
temporal
coherence
region.
Van Noorden made
quantitative
measurementsof
the
fission
boundary(see Figure 9).
For
tempi
of about
2.5
to 10
tones/sec.,
the
fission
boundary
is more or less
hori-
zonal,
i.e.,
the
intensity
difference
(AL) necessary
for a
segregatedpercept
does
not
change
with
tempo
over this
range,
but
lies about 2 to
4
db
on either
side of
the
referenc
tone level. For
tempi
less than 2.5
tones/sec.,
the minimum
level
difference
for
fission increaseswith
decreasing empo
(or
longer
inter-tone
intervals
of
silence)
and is
symmetric
about AL
=
0
(no
difference
in
level).
For
tempi
greater
than
10
tones/sec.
the level differenceat which fission occur
increases
with
increasing
empo
but
the situation s
not
sym
Tempo
(Tones/sec)
20
0
5
3 2
Masking
SContinuity
"
LA>
LB
20o
Fission
LA
>LB
08
.,
Roll
Inevitable
ev
le
--*
35 db
SP
_Coherence
20
Fission
LA
-
0
0.
mtTe
r
?
Time
Time
-
Figure
21. One result
of
McAdams'
tudy
(1977)
suggested
hat there
is
an
interaction
between
pitch "height"
and timbral
"sha
ness"
(see
text).
In
a
repeating
4-tone
sequence,
one of the
pairs
of
tones was
selectively
enriched
by adding
he
third
harmonic
A
greaterdegree
of
segregation
of
the
high
and low
streams
was found for the
formercase.
The
dashed ines
indicatethe two-
stream
percepts
and the
dotted
lines
indicate
a
potential
one-stream
percept.
The
vertical
solid lines
represent
he fusion and
timbre
of
the 2-tone
complex.
Page
38
Computer
Music
Journal,
Box
E,
Menlo
Park,
CA
94025
Volume 3
Numb
-
7/27/2019 McAdams Bregman Streams
15/21
(or
frequency) organizations
and
simultaneous
(or
timbral)
organizations
n the
formation
of
auditory
streams.
The
stimulus used
by
Bregman
nd
Pinker
was a sine
tone
alter-
nating
with
a
two-tone
complex.
In
Figure
22,
tones
A and
B would
represent
he
sequential
organization
and
tones B
and
C would
represent
the
simultaneous
organization.
The
harmonicity
and
synchronicity
of
tones B
and
C
in
the
complex
were
variedaswas the
frequency separation
between
the sine
tone
A
and the
upper
component
B
of
the
complex.
The rationale or
varying
hese
two
parameters
was as
follows.
Tones with
frequency
relationships
derived
from
simple
ratios,i.e. those that exhibit "consonance,"should tend to
fuse
more
readily
than
combinations
considered
to
be
dis-
sonant. While
the
evidence for
this
was
very
weak
in the
Bregman
and
Pinker
study,
work
currently
n
progress
n
Bregman's
aboratory
strongly
suggests
that
this
is
indeed
the
case.
The
new
evidence
further
suggests
that
the
effect
of
harmonicity
tself
is
relatively
weak
and
may
be
over-
ridden
by
stronger
factors
such
as
frequency
contiguity
and
synchronicity
of
attack
and
decay.
Tones
with
synchronous
and
identically
shaped
attack
and
decay
ramps
are
more
likely
to
fuse
than
those
with
asynchronous
or
dissimilar
attacks
and
decays
[12, 15].
This
may
be
a
major
cue in
being
able
to
parse
out the
different
instruments
playing
together
in
an
orchestrasince they all have substantially
different attack characteristics.
n
addition,
there is
a
very
low
probability
of several
people precisely
synchronizing
he
attacks.
In
light
of this
work,
one
might
make the
following
predictions
or the
perception
of the stimuli
used
by
Bregm
and Pinker:
1) Sequential streaming
s favored
by
the
frequency
proximity
of tones
A and B
(as
we have
illustrate
in the
earlier
examples).
2)
The
simultaneous
or timbral)
fusion of tones
B and C is favored
by
the
synchrony
of
their
attacks.
3)
These two effects
"compete"
for
tone B's
memb
ship
in their
respective
perceptual
organizations.
4)
Finally
when tone B is "captured"by tone A, it is removed romth
timbral
structureand
tone C sounds less rich.
Thus,
it
is
reasoned hat if
the two
simultaneous ine
tones B and
C are
perceived
as
belonging
to
separate
streams,
they
should
be
heardas sine tones.
But if
they
are heard
as one
stream,
hey
should
sound like one
rich
tone.
It would
be
appropriate
o
introduce
the notion
of
"belongingness"
t this
point,
since
we talk of tones
belongi
to
streams,
and
of
frequency components
and the
timbre
resulting
rom their interaction
belonging
o a
perceived
on
event.
"Belongingness"
a
term used
in the
perceptual
itera
ture of Gestalt
psychology)
may
be
consideredas
a
principl
of
sensory organization
which serves to
reconstruct
physic
"units" nto perceptualeventsby grouping ensoryattribute
Stimulus
C B
LL
t
4-
Time
-
Percepts
:3
A
A
Cr
B
B
\
,
B B
C C C C
A &
B
Stream
A
&
B
Segregate
C
Pure
C
Rich
Time
-
Time
-
Figure
22.
The
competition
between
sequential
and simultaneous
organizations
n
the
formation
of
auditory
streams
s
shown
here.
Tone B
can
belong
either
to the
sequential
organization
with tone
A or to
the simultaneous
organization
with tone
C
but not to both at
the
same time.
Bregman
and Pinker
(1978)
varied
the
frequency
separations
between tones
A and
B
an
between
tones
B
and
C and also varied he relative
synchrony
of onset of tones B
and
C.
The
dotted
lines
in the
figure
indicat
the
stream
percepts
and the
vertical
solid
lines
represent
he fusion
of
tones
B and
C
(cf.
Taped
Illustration
15).
Stephen
McAdams
nd
Albert
Bregman:
Hearing
Musical treams
Pag
-
7/27/2019 McAdams Bregman Streams
16/21
of
thoseevents nto
unified
percepts.
As
Bregman2]
points
out,
"belongingness
s a
necessary
utcomeof
any
process
which
decomposes
mixtures,
ince
any
sensory
ffect
must
be
assigned
o some
particular
ource." n
this
case,
when
the
simultaneous
ones
B
and
C are
segregated,
he
timbre
resulting
rom
heir
nteraction
till
exists
andcanbe
heard
f
one
istens
or
t,
but it
is not
perceptually
ssigned
o
either
of
the tones
B
or
C,
and
husdoes
not
affect he
perception
of them.The
nature
f a stream
s
such
hat ts
qualities
re
dueto the
perceptual
eatures
ssigned
o
it.
In
Taped
llustration
5
one
can
heara
case
n
which
A
is close o BinfrequencyndCisasynchronousith B.Then
a
case s
heard
where
A
is further
way
rom
B in
frequency
and
C
is
synchronous
ith
B.
Tones
B
and
C
have
he
same
frequencies
n
both
cases.Listen
or
both the
A-B
stream
nd
the richness f
tone
C.
The
istenersn
Bregman
nd
Pinker's
study
reported
erceiving
as
being
icher
when
B
and
C
were
synchronous,
nd
this
judged
ichness
ropped
ff
with an
increase
n
asynchrony,
.e.
as
C either
preceded
r
followed
B
by
29
or 58
msec.
As
the
frequency
eparation
etween
A and
B was
ncreased,
was
reportedly
erceived
s
being
increasingly
ich.
The
Role
of
Context
in
Determining
Timbre
These
indings
ndicate hat
the
perceived
omplexity
of a
moment
of
sound s
context-dependent
see [19]
as
another
xample
of
the
trend
toward
viewing
imbre
as
depending
n
context).
Context
may
be
supplied
y
a
number
of
alternative
rganizations
hat
compete
or
membership
f
elements
ot
yet
assigned.
imbres a
perceived
roperty f
a
stream
rganization
ather han
he
direct
esult
f
a
particular
waveform,
nd
is
thus
context-dependent.
n
other
words,
two
frequency
omponents
hose
synchronous
nd
harmonic
relationships
ould
cause
hem
o
fuse
under
solated
ondi-
tions
may
be
perceived
s
separate
ine
tones if
another
organizationresents
tronger
vidence
hat
they
belong
o
separate
equential
treams.
A
very
compelling
emonstration
f the
decomposit
of a timbre
organization
y
alternate
requency treamin
organizations
s illustrated
n
Figure
3. When
one A
is
presented
y
itself
t elicitsa
timbre,
enoted
TA.
f this
ton
is
preceded y
tone
B
eliciting
imbre
TB,
andsucceeded
y
tone
C
eliciting
imbre
TC,
one noticesthat
timbre
TA
completely
isappears
nd s
replaced y
timbres
TB
and
TC
Here
the
highest
and lowest
components
f tone
A
are
streamed
with those
of toneB and
subsequently
ssume
timbre
dentical
o thatof toneB.
Also,
he inner
ompone
of tone A
streamwiththose of tone
C andassume
like
timbreTaped llustration6).
An
important
uestion oncerning
he
assignment
f
timbre
and
pitch
to a tonaleventarises.
Bothtimbre
and
pitch
havebeen found
to be context
dependent10,
21,
respectively].
But each
may
be determined
y
different
ongoing
contextual-organizations;
s such
they may
be
considered
o
be
associated
erceptual
imensions
f a
soun
but
may
not be
inextricably
ound
o one another.How
rel
vant o music
heory,
hen,
arestudies hat
dealwiththe
pe
ceived
pitch
and imbre
f tones n isolation?
This
question
not meant o insinuate
hat
sensory
nd
psychophysicalxp
imentation reuseless.
Far rom t. If we
think n termsof
investigating
he
experience
f music
at
different
evels
of
processingFigure 4), we seehow important ll of these
areas f research
re n
building
he whole
picture.
The
raw
physical
nput
s
modified
by
the
limitsof the
sense
organ
whose
output
s stillfurthermodified
y
cognitive rocesse
But
by
studying teady-state,
r at least
relatively
imple
signals,
we can find
the limitsand
interactions f the
sens
organs
nd
peripheralrocesses,
uch
as
temporal
nd
spect
resolution,
ateral
nhibition,
nd
masking,
hich imits
affe
the final
percept.
Beyond
hese,
the
central
perceptual
processes
uch as
pitch
extraction,
imbre
buildup,
and
coherence
nd fission
modify
the initialneural
result
of
stimulation
f the
sensory
ystem.
Further
nteractions
n
higher
rain
processes
uchas
attentional
rocesses,memor
and
comparison
f
pitch,
imbre nd
oudness,
ontext
extr
One
Timbre
Two
Timbres
:TA
versus
I I :
1 2
11
A
B
A
C
Figure
23.
The
first
part
of
this
figure
shows
a
repeating
one
(A)
consisting
of
4
harmonics.
This
tone would
elicit
a certain
timbre
percept,
TA.
In
the
second
part
this
tone
is
preceded
by
tone
B,
consisting
of the
top
and bottom
harmonics,
and is
suc-
ceeded
by
tone C
consisting
of the
two
inner
harmonics.Tone
B
elicits timbre
TB
and
tone C
elicits
timbre
TC.
However,
due to
the
streaming
of tone
A's
components
with
those of
tones B
and
C,
TA
totally
disappears
nd
is
replacedby
TB
and
TC
(cf.
Taped
Illustration
16).
Page
40
Computer
Music
Journal,
Box
E,
Menlo
Park,
CA
94025
Volume3 Numb
-
7/27/2019 McAdams Bregman Streams
17/21
Physical
Environment
Levels
of
Processing
Temporal
and
Spectral
Resolution
Lateral
nhibition
"Sensory"
Masking
etc.
Pitch
Extraction
Timbre
Buildup
"Perceptual"
Coherenceand FissionLimits
etc.
Attention
Memory
Context
Extraction
Form andTextureIntegration
etc.
Figure
24.
This block
diagram
uggests
a
possible
arrangement
of
the
processing
of
acoustic
information
at
different
inter-
connected levels
of
the
auditory
system.
tion,and ormand exturentegration,culpt hetransduced
information
nto
meaningful ercepts.
t is felt
thatall
of
these
levels
of
processing
eed
into each
other n a
sort
of
heter-
archical
as opposed
o
a
hierarchical)
ystem.
The
point
being
made
s that
in
the
framework
f
music
where
all of
these
complex
nteractionsre of
great
mportance,
he
context
that
s created
may
be
the
essential
eterminantf
the
musical
resultof
a
given
ound.
One
sound s
potentially erceivable
n
a
great
number
f
ways,
depending
n its
context.
Melody
This eads
us
to
believe
hat
the
fundamental
erceptual
element n musicmaybe the "melody" ather hanthe
isolated
one.
Or
n the
terminology
f
auditory
erception,
the fundamentaltructure
s the
auditory
tream.
This s
not,
of
course,
new
notion;
but an
empirical
pproachmay
allow
us to
clarify
and
delimit he
concept
o the
extent
that
we
may
predict
he
perceptual
esults.
That,
n
themindof the
first
author,
s the
primary
oncern f the
composer.
et
us,
then,
examine
melody
and ts
relation o
attention.
For our
purposes,
we can
think
of
melody
as a
connected nd
ordered uccession
f
tones
[28].
It
follows
thenthat
temporal
oherence
s
necessary
or
a
sequence
f
tones
to be
perceived
s
a
whole.
On the
other
hand,
a
sequence
f tones
may
segregate
nto two
or more
separa
streams
which
are
ndividually
oherent.
n
this
case,
we wo
perceive
everal imultaneousmelodiesrather
han
one.
If
such
an
operational
efinition
f
melody
s
tenable
we
must hen
question
ow t is thatsome
of the extension
elements
ther han
pitch
for
"melodic"
material
revali
perceptually.
or
example,
when
a
composer
ses timbre
thematic
material,
o
we
still
perceive
he
sequence
s main
taining
temporal ontinuity?
ometimes
oherence
s
mai
tained
by
sensitive
erformers
ndsometimes
t can
be
very
difficult
o
perceive.
Of
course,
we
have
not
included
th
elementswhichaffectthe perception f melody,suchas
underlying
armonic
tructure,
ut
we
are
only
attemptin
convey
the notion that
temporal
coherence
hould
be
considered
ssential
o
melody
ormation.
Conversely,
ne
may
use the
principles
f
fission
o
develop
ules
or
creati
polyphony
nd
counterpoint
n
sequences
f
acoustic
ven
Attention
ndMusical
tructure
It
has
become
apparent
o
the
first
author
hatmusic
structure,
sit is
perceived
n
real-time,
s
inextricably
ou
to
attentional
rocesses.
bit
of
introspection
illreveal h
there
are at least
two
kinds
of
attentional
rocesses,
whic
we
might
allactiveor willful
attention,
nd
passive
r
auto
matic
attention.One
mightwillfully
direct
one's
attention
some
object
or
sequence
of
events,
such
as
listening
o
particular
vents
within
a
piece
of
music.
Or,
someunusu
event
might
attract
ne's
attention
nexpectedly,
uchas
th
honking
orn
of
an
oncoming
ar
you
had
not
noticedas
yo
stepped
ntothe streetabsorbed
eep
n
thought.
Thathorn
demands
our
attentionand
n
all
probability
ets
t in a
hurry.
n
particular,
othkinds
of
attention
mayparticipa
in the
process
f
listening
o
auditory
treams.
or
nstance
in
Figure
when
a
sequence
f tones
ies
above
he
tempor
coherence
boundary,
o amount
of
activeattention
can
extract
he
percept
f
one
coherent tream.
Here
perceptio
is limited
by
passive
ttentional
rocesses 28].
Thishas
importantonsequencesorcomposers ho ntend o usefa
melodic
equences,
ince
t
suggests
hat there
are
tempi
a
which
he
listener
may
not
be able o
follow
as
a
melody
h
sequence
ou
have
constructed,egardless
f the attention
will
power
nvoked.An
example
f theseeffects
may
be
fou
in
the
sequences
resented
t different
ates n
Charles
od
Earths
Magnetic
Fields.
Without
pretending
o
know
the
composer's
ntent,
t can be
amusing
o listento
the same
sequence
decompose
nd
re-integrate
tself
during
variou
tempochanges.
n
a multi-streamed
equence
ne can rela
attentional
effort with the
result
that
attention
might
randomly
alternate
among
the available
treams.Or one
mig
selectively
focus
attention on
any
one of them
individual
and evenplay them againstone another.
Van Noorden
reported
that the
temporal
coherenc
boundary
(the
boundary
below which
all tones
may
belon
to
one
stream)
s not affected
by
previousknowledge
of the
sequence,
and considered
t to be a
function
of a
passive
attentional
mechanism,
given
that the listener
"wants to
hea
coherence."
However,
what
happens
between
the boundari
of
temporal
coherence
and fission
dependsupon
a number
factors,
such
as
context,
and seems
to
be under
the influenc
of attention.
Dowling [17],
for
example, reported
that if
listeners knew beforehand
whi:h
melodies were
being
inte
leaved,
they
could,
with
a
bit of
practice,
extract the
appro
Stephen
McAdams
nindAlh"rt
Rrtnmn-
o
Ra,;
...A--
Pag
-
7/27/2019 McAdams Bregman Streams
18/21
priate melody
even at
very
small
separations
of
the
ranges
traversed
by
each
melody.
It is
currently
assumed
that this
ability
would
degenerate
at faster
tempi.
In
addition,
van
Noorden
found an
effect
at
very
fast
tempi
where t
was
virtually
mpossible
o
hear
sequences
within a
range
of two
or three
semitones as
other than
temporally
coherent.
At
very
fast
tempi
(about
12.5
tones/
sec.)
the
tones
of
such
narrow
patterns
are not heard
as
separate
membersof a
sequence,
but
actually
merge
nto
a
continuously
rippling
exture,
as one can
hear in
Taped
Illustration
17.
There
are
thus
attentional
limits
in
the
ability
of
the
auditory system
to tracka
sequence
of events. When events
occur too
quickly
in
succession,
the
systemusesthe variousorganizational ulesdiscussed n this
article
to
reorganize
he
events
into
smaller
groups.
It
may
then track
events
within
a
particular
group
if
the
listener
is
paying
attention
to
it,
but
this
narrowing
f
focus
necessarily
causesa
loss of
information.One result s
the
inability
to
make
fine
temporal
order
udgments
between
streams.
These
organi-
zational mechanisms eflect the
tendency
of the
auditory
sys-
tem to
simplify things
in
the face of
excessive
complexity.
In
the
example
where the
fast
sequence
of
tones
merges
nto a
continuous
"ripple,"
he
auditory system
is unable to
success-
fully
integrate
all of
the
incoming
nformation nto a
temporal
structureand
simplifies
the
situation
by
interpreting
t
as
texture
(see
also
[31]
).
Thus
the
auditory
system,
beyond
certaintempi,mayinterpret he sequenceasa singleevent and
assign
o it the
texture
or timbre
created
by
its
spectral
and
temporal
characteristics.
An
understanding
f
(or
intuition
about)
these
organi-
zational
processes
can
lead to new
dimensions
of
control
over
musical
structure.
For
example,
one
might
construct contra-
puntal
sequences
that
play
across various
stream
boundaries
and
through
different
borderline
regions
between
temporal
coherence
and
fission.
(It
may
be that
composers
such
as Bach
were
already
using perceptual
ambiguityconsciously
in their
work.)
Any
or
all
of
the
relevant
musical
parametersmight
be
used
to
accomplish
this.
Then,
an
appropriate
use of events
that
vie for or demand he
listener's
attention
can be
used
by
the
composer
to
"sculpt"
the
attentional
processes
of the
listener.Sincesomeeventsseemmore
striking
o some
persons
than to
others,
this
attentional
sculpture
n
time would lead
different
listeners
through
different
paths
of
auditory
experience.
Further,
perception
s
bound
to
vary
from
time
to
time within
a
single
person,
so
the
experience
would
be
different
with
each
listening.
A
composition
of
sufficient,
controlled
complexity
might
thus be
perceptually
nfinite
for
a
given
istener.
Conclusion
An
attempt
has
been
made here to
point
out that
composers
and
music theorists should
thoroughly
examine the
relationshipbetween the "musical"principlesthey use and
espouse,
and
the
principles
of
sensory,
perceptual,
and
cognitive organization
hat
operate
in the human
auditory
system. Many
of
the
principles
discussed n this
article extend
to
higher-level
perceptual
analysis
of
musical
context
and
structureand
may
well
represent
a
scientific
counterpart
o
some
extant music
theoretical
principles.
In other
cases,
though,
this
group
of
phenomenasuggests
perceptualorgani-
zations
which have
little
relation to
methods
currently
used
to construct
or
analyze
musical
structure. To
ignore
the
evidence
from
the real
life
system
in
developing
a
theory
of
music
or a musical
omposition
s to takethe chance
of
relegating
ne's
work o the realm f what
might
be
termed
"paper
music."
Acknowledgments
This
paper
s basedon a
workshop
ntitled
"The
Perceptual
actoring
f
Acoustic
Sequences
nto
Musica
Streams"
elivered
t the
1978 International
omputer
Music
Conference,
vanston,
llinois,
and
published
n
the
Proceedings
f
the
Conference.
he
authors
would
ike
to
thank
Dr. Leon
van
Noorden or
his
many
helpful
uggestio
andvaluable riticisms f themanuscript;igures , 8, 9,
10
and
20 were
redrawn,
with
permission,
rom
Dr. van
Noorden'shesis.
The
present
version f this artic
was
prepared
hileMr.
McAdams
as
a Graduate
ellow
of the
National
ScienceFoundation.
Dr.
Bregman's
esear
hasbeen
supported
y
grants
rom
he
NationalResearch
Council
f
Canada,
he
Quebec
Ministry
f
Education,
nd
the
McGill
University aculty
of
Graduate
tudies
and
Research.Muchof
the research sed
the
facilities f the
Computer-Based
aboratory
f the
McGill
University
Department
f
Psychology.
The
editors
of
Computer
Musi
Journalwould ike
to
thank
Mr.
McAdams
or
preparing
he
illustrations
n
this
article.
Appendix
Description
f
Taped
llustrations
1.
A
repeatingequence
f
three
high
ones
1600,
2000,
2500
Hz.)
s
interspersed
ith three
ow
tones
350,
430,
55
Hz.)
used
by
Bregman
nd
Campbell
1971). a)
At
a
tempo
o
five
tones/sec.
he
sequence
s
perceived
s one
stream f
alternating
igh
and ow tones
(cf.
Figure a).
b)
At
a
tempo
of
ten
tones/sec.
he
sequence
egregates
erceptually
nto
on
streamof
high
tones and one stream
of
low tones
(cf.
Figure
b).
2. Another
epeating
ix-tone
sequence
s
played
at
a
tempo
of ten
tones/sec.,
but the
higher
riplet
s closer n
frequency
o the
lower.Tone
F
may
be
perceived
s
belong
to either
he
high
stream r the
low
stream
epending
n the
listener's
ocus.
Note
that tone
F
cannot
belong
o both
streams t
once
(cf.
Figure
).
3.
A
repeating
ix-tone
sequence
tartsat
a
slow
tempo
As the
tempo
s
gradually
ncreased,
he
sequence
s
progres
sively
decomposed
nto
smaller
erceptual
treams
ntil
t is
no
longerpossible
o
follow
he tonaleventswhich
merge
n
the
percept
f
timbre
r
texture
cf
Figure ).
4.
Using
he same nitial
equence
s
n
Taped
llustratio
3,
the
frequencyeparation
etween
emporallydjacent
on
is
gradually
ncreased. he samesort of
decomposition
nto
smaller streams occurs. The limits in this example are
determined
ot
by temporal
esolution
ut
by
the
audible
frequency
ange
cf
Figure
).
5. The
tones of two familiarmelodies are interleaved n
th
same
frequency