Week 10

Question 0

Attention does not permit us to see two things simultaneous

In music

Answer 0

Can’t hear two themes at once.
Access schemata, cannot attend to all perceptions
Memory, schemata, expectation, attention
One dependant on the other.
Constantly generating expectations? Wouldn’t survive without it.
When, wron, comes to attention.
Our notions of attention, memory, expectation, and schemata are
highly intertwined. Without attention, which allows us to selectively
encode certain important streams of perceptual information from the
“blooming and buzzing confusion of the environment,” we could not
store perceptual information in any efficient way. Without memory, we could not generate expectations for the future, since expectation is
based on the projection of prior experience onto the future. Without
expectations, there cannot be attention, since it is our expectations
that help to direct attention towards “the right spot.” What is more,
nearly all of these processes depend on learned and/or innate
schemata, e.g. that of a dalmatian, which help in “denoising” a blurry
perceptual picture and yield more meaningful perceptual information.

Question 1

Krumhansal serial music

Answer 1

Non musicians, recency effect happened.
Similarity paradigm.do we listen through a tonal filter, yes some notes got bumps becuase of this. Suggesting local key.

Musicians, did EXACTLY THE OPPOSITE

Knew should not hear repeated note. Knowing structure. Rate it low if hear same tone.
Did also seem to use tonal filter. High fit with note far from suggested keys,

Question 2

Characteristics of schemata

Answer 2

They embody general knowledge of stimulus
properties of a given object.
In music:
▫ Pitch
▫ Loudness
▫ Timbre
▫ Spatial Location
Pentatonic
Schema
Clapping
• They are more general than the sounds that are
actually heard.
• They direct our attention, interact and interfere
with memory, and guide our expectations of what will happen next.
Stairs at art museum note.

Question 3

Schema types

Answer 3

PET
Prototypes – Abstracting the common features of
similar experiences to create a generalized
experience (statistical learning)
E.g. Cadences, scales

 Exemplars – Well-learned individual cases. Schema for one.
e.g. Beethoven’s 9th Symphony

Theories – Theories derived from past
experience. Deduct, availability heuristic.
E.g.Saxophone = Jazz

A schema is a shorthand for a packet of knowledge, be it an abstracted prototype, a well-learned
exemplar, or a theory intuited from experience.

Question 4

Attention

Answer 4

The processing of sensory information into a percept is highly
influenced by attention. Features of a stimulus that are perceptually
salience will compel us to pay close attention. At the same time, to a
certain extent we can also consciously direct our attention towards
features that might not be perceptually salient at all. With some
amount of training, for example, we can hear out partials of an
otherwise fused complex tone.
Conscious, or unconscious attention. Cocktail party in music.

The cocktail party effect is an important phenomenon in the study of
hearing. Despite highly complex and confusing acoustic
environments, such as may be found in a cocktail party, people can
allocate attention towards their conversational partner and understandwhat they say. Music could therefore be understood as a melodic cocktail party, since in music we also have multiple simultaneous and sequential streams that are competing for the listener’s attention.

Question 5

Attending to melodies

Answer 5

How do we parse out voices/streams from the musical cocktail party?
Although we can selectively attend to either melody based on the musical features
distinguishing the two melodies from one another (as a result of simultaneous and
sequential grouping cues from auditory scene analysis), we cannot attend to both at
the same time.

In 5.1a, the melodies are distinguished by register (pitch height), so it’s easy to
selectively attend to either melody.

In 5.1b, the melodies are distinguished by timbre.

In 5.1c, they are distinguished by loudness, and in 5.1d, by spatial position.

In 5.1e, Dowling presents two interleaved melodies, “Mary Had a Little Lamb” and
“Three Blind Mice.” With no differences in any musical features to distinguish the twomelodies, it is very difficult to hear either one without first hearing the melodies in
isolation. Thus, in 5.1e you do not know which schema to use and so cannot easily discern the tunes.

Question 6

Memory

Answer 6

Memory is at the basis of any sort of knowledge. Memory involves processes of encoding, storage, and retrieval.

There are numerous proposed memory systems:
Long-term Memory (Weeks, Months, Years…)
 Semantic – general knowledge structures
 Episodic – personal experiences
 Declarative – explicit recall (knowing that)
 Procedural – implicit retrieval (knowing how)

Short-term Memory (Up to 30 seconds)
 Working – Permits conscious access, internal rehearsal
 Sensory – a short echo of the past stimulus

Memory is one of the most important concepts in cognitive psychology. One persistent approach towards researching memory is to identify and characterize
its different subsystems. Instead of one unified model of memory, it seems that
human memory can be understood as comprising multiple parts.

Long term memory preserves stored items for long periods of time. One
distinction is between semantic and episodic memory. The latter stores traces
related to personal experience (“I went to hospital in March.”), the former
responsible for general knowledge structures (“Australia is not close to Austria.”).Declarative long term memory denotes memories that can be explicitly recalled.
Procedural long-term memory refers to processes stored implicitly, as e.g. the
motor memory a saxophonist might possess for how to play a certain tune,
without being able to explicitly write down the melody (“know that“ vs. “know
how”).

Short term memory is thought to last for up to 30 seconds in audition. Here,
working memory is differentiated from sensory memory. Sensory memory
behaves like a short lasting echo of past stimuli and thus can easily be
overwritten with novel incoming stimuli. Working memory allows for more
conscious access, processing and internal rehearsal. The task of singing a
melody you just heard backwards, is one concerning auditory working memory,
for instance.

Question 7

Pitch memory

Answer 7

Krumhansl investigated how musical schemas like scales affect
memory for pitch. Listeners were presented with either a pitch found in
the key of C (diatonic), or not found in the key of C (nondiatonic), then a
series of distracter tones in the key of C (tonal), or not in any key (atonal).They then heard the comparison tone, which was either the same as the standard or one semitone away, and had to indicate whether the comparison tone was the same tone (i.e., a same-different task).
Results indicated that memory was best when the comparison tone
was a diatonic member of the tonal interference scale. If it was non-
diatonic, performance degraded. In other words, if the listener is trying
to remember a standard pitch as a chroma in a particular key (a scale
degree), the atonal context hurts performance. If the listener is trying to
remember a pitch foreign to a key, then a tonal context in C is disruptive
and an atonal context is not. These results indicate that schemata play
a major role in memory for pitch structures. If listeners have
incorporated schemata like that of tonal, diatonic music, they will havea much easier time remembering pitches and melodies that follow the schema.
Classic interaction.

Question 8

Melody recognition

Memory for melodic feature

Answer 8

In addition to chroma, other important cues exist. The most importantmight be contour – the overall “up-and-down” shape of the melody.
Listeners heard pairs of atonal melodies, the original on the left, and then
the comparison melody, which was either an exact transposition (A), a
melody with different intervals but the same contour (B), or a melody with a different contour (C). Listeners easily distinguished between A-C and B-C (85-90%), but couldn’t distinguish between A-B (chance level). Since
discrimination was around chance for melodies altered in intervals
only, memory for atonal melodies must rely particularly on contour
information.
In this early experiment on memory for melodies, Deutsch played
highly familiar melodies in octave scrambled form.

5.3a preserves abstract chroma (scale degree), but eliminates both intervalsize and contour. Listeners recognized these melodies with 65% accuracy.
(“Mary Had a Little Lamb”)

5.4b does not preserve interval size, contour, or abstract chroma, and at
this point listeners recognize these melodies with 10% accuracy.

Question 9

Expectation

Answer 9

Knowledge-based process that prepares the organism to
better react to future events.
Whereas memory helps to retain a sense of the past,
expectation allows us to generate a sense of the future.
In music psychology: Fulfilled or violated expectations formthe basis of the experience of surprise, tension, pleasure
and meaning in music (Meyer 1956, Huron 2006).

Conscious expectations are also possible
There are many types of expectations:
 What? (the expected event)
 When? (the expected time of event onset)

Sensory
 generated by the sensory similarity between events.Schematic

relating to distinct musical schemata/styles/genres (e.g., a cadence, sonata form, etc.)
Veridical

relating to a single musical piece/passage (e.g., the
beginning motive of Beethoven’s fifth symphony)
Dynamic
 being generated in the course of listening
independent of earlier knowledge of the piece

Question 10

Memory for Pitch – Summary

Answer 10

 Short-term memory for pitch and melody
subjected to interference are dependent on:  Time interval
 Similarity of interference tones to standard Tonal/atonal scale schema
 Contour
 Long-term memory for melodies is dependent on:
 Abstract Chroma (scale degree)
 Contour

Question 11

Measuring expectancy

Answer 11

Direct: How much did you expect this tone?/ How
well does it fit?

Problem: Attention towards task might distort the
phenomenon.
Indirect: Exploit behavioural manifestations of
expectancy:
We are faster and more accurate at processing a
stimulus when it is expected, or primed.


Define a task (on which listeners can be faster/slower, more/less accurate) to probe expectancy without
making subjects aware of the aim of the study

However, the effect might also 
be due to sensory structure of 
the employed timbres 
A computational model of 
sensory tonal contextuality wasused to simulate this theory. 
Results indicated that for piano timbres used in Exp. 1, this 
interpretation cannot be ruled 
out. 
The bottom figure shows 
context effects for sine tones. 

Primarily cognitive schemata shape our expectations of future tones in memory, thoug. sensory expectations may also play a role.

In priming experiments one usually tests the effects of slight
alterations of context on the perceptual processing of a target item. Inthis experiment, one note (and its repetitions) were changed in the
first measure of a set of two-bar melodies. This turned the last tone
from a tonic into a subdominant (which the authors surmise is less
expected in this context).
The participants task was to differentiate the timbre of the last target
tone. Note that this task is not directly related to the expectedness of
the note itself. However, it serves as an indirect measure by
evaluating correct responses and reaction time.

Question 24

Expectation

4 main types.

Answer 24

There are many types of expectations:
 What? (the expected event)
 When? (the expected time of event onset)

Sensory
 generated by the sensory similarity between events.Schematic

relating to distinct musical schemata/styles/genres (e.g., a cadence, sonata form, etc.)
Veridical

relating to a single musical piece/passage (e.g., the
beginning motive of Beethoven’s fifth symphony)
Dynamic
 being generated in the course of listening
independent of earlier knowledge of the piece

Question 25

Cognitive representation, hierarchical parts of music the air, how do we hear it in time.
Abstract knowledge distinguish
Dedi

Answer 25

Abstract knowledge structures
Event structures
implicit knowledge about the structure mental organization of musical events of music acquired through exposure in
culture

2 types of abstractions: 
• systems of relations among musical 
categories 
• lexicon of abstract, archetypical 
patterns or idioms 

atemporaof a given piece real-time processing of musical eventsand patterns within the context of a learned system of relations (abstract
knowledge structures) ongoing temporal sequence of events

The incoming acoustic information is parsed and interpreted according to acquired
abstract knowledge structures which affect the subsequent encoding and organizing
of the musical material in accumulated event structures. Abstract knowledge
structures, such as the tonal hierarchy for example, refer to the rules by which pitchesare ordered in terms of dominance and stability in a scale structure. Event structures
involve a mental organization specific to a piece of music, and they depend on
grouping structure, metric structure and abstract knowledge structures that establish patterns of tension and relaxation.

Question 26

Event structure processing hypothesis

for form-bearing dimensions

Answer 26

• Relations among event attributes should be able to be processed through time in order to serve inthe encoding and organizing of musical material in an accumulating mental structure that
represents the momentary perceptual
comprehension of the musical form.

Schematic diagram of musical (event 
structure) processing – Lerdahl & 
Jackendoff (1983) 
event structure 
hierarchy  
network  
of tension and relaxation 
Grouping 
structure 
Stimulus 
Metric 
structure 
Time span 
segmentation 
Time span 
reduction 
Prolongational 
reduction 
Stability 
conditions 
abstract knowledge  

Lerdahl & Jackendoff’s (1983) Generative Theory of Tonal Music (GTTM) was a
seminal work in the field of cognitive music theory by building a theory whose many
components could be tested experimentally. The GTTM schematic diagram of musical processing contains a chronological framework relating to event structure processing. The musical stimulus is analyzed by processes of grouping and metrical organization,
which determine the “time span segmentation”. Event structures are hierarchically
organized in a “time span reduction,” based on grouping processes of segmentation
and stability conditions (abstract knowledge structures, such as Krumhansl’s tonal
hierarchy). Finally, the “prolongational reduction” stage is based on a hierarchical
network of tension and relaxation. We will discuss the timespan segmentation, the
timespan reduction (related to event structure hierarchy) and the prolongational
reduction (related to patterns of tension and relaxation) today.

Question 27

According to the GTTM, grouping structure should reflect the natural perceptual
grouping of events into “chunks” that make sense musically. Because of principles
such as duration contrast, silences, articulation, and parallelism (pattern repetition), example (a) is a bad grouping structure and (b) is a good one. This can be verified byintroducing pauses at group boundaries. They sound OK for a good grouping and
awkward for a bad grouping.

Answer 27

Grouping derives its principles from Gestalt psychology, most of the principles of which stem from the principles of spatial proximity (temporal in music) and formal similarity in vision.
Examples 3.6 and 3.7 show the proximity principle in vision and its translation into music in 3.8. Example 3.9 shows the similarity principle in vision and its translation into music in 3.10 and 3.11. Note that spatial proximity in vision translates into temporal proximity in music. Formal similarity in vision translates into similarity in auditory attributes such as pitch, loudness and timbre in music.
3.6 demonstrates that proximity makes a 2/1 grouping in (a) and a 1/2 grouping in
(b).
3.7 shows a strong grouping in (a), a weaker grouping in (b) and no grouping in (c)
due to the equal spacing of the circles.
3.8 is a musical translation of these principles.
3.9 shows a 3/2 or a 2/3 grouping on the basis of formal similarity (square vs. circle). 3.10 shows the same principle with a pitch difference.
3.11 shows the relative degree of grouping strength created with temporal proximityin 3.8 with relative pitch similarity.

Question 28

Lerdahl & Jackendoff proposed a number of grouping preference rules that
structure musical sequences. Two principals underlie the rules: proximity
(relating to temporal events where the boundary is set after the ends of slurs or rests or after a prolonged sound among other short ones) and similarity (so-called “change rules” where the similarity of events is registered after hearing the first differentiating element initiating the following group). The rules result
from a change or discontinuity in duration (1, 2, 5 and 6) or auditory attribute (3, 4, 7). This figure represents Deliège’s (1987) conceptualization of Lerdahl and Jackendoff’s rules.
Who tested this and how?

Answer 28

Note acoustic and duration differences
Deliège tested the principles put forth by Lerdahl & Jackendoff with musical examplesutilizing the different grouping principles. In Experiment 1, listeners were asked to
indicate when they heard grouping boundaries and then these were analyzed in
terms of whether they fit with the principles outlined by Lerdahl & Jackendoff. Most
of the segmentations were in accordance with the rules, and they were more in
accordance with the rules the more musical training the listeners had. 4 levels of musicians
Non musicians seem to follow rules only half the time.

Musical parameters most imp.
This histogram shows the percentage of responses in accordance with the rules for
each principle. Attack point and dynamics were most often used by listeners.
Note that in general musicians (hashed bars) made segmentations that were more inaccordance with the rules than did nonmusicians (black bars). Nonmusicians were
less sensitive to principles such as slur/rest, articulation and note duration (legato,
détaché, staccato).

Question 29

Conflicting organization principles

Answer 29

Grouping rules according to different principles can enter into conflict with one
another in real music. This also occurs in other sensory systems, such as vision as
shown in Example 3.12. In (a), the proximity and similarity rules converge on a 2/3
grouping. In (c), proximity wins out over similarity, whereas in (b) the grouping is
more ambiguous, although one can see either 2/3 on the basis of similarity or 3/2 onthe basis of proximity. Example 3.13 shows a musical translation of this using time
and pitch.

The results show that people had a slight tendency to prefer the change that
occurred in the middle of the sequence (4 and 5 are higher than 3 and 6). Register,
dynamics and timbre are the strongest dimensions for signaling groupings against theothers, although Deliège did not really control for equal perceptual change along
each dimension.

Question 30

Time span reduction refers to?

Answer 30

Event structure processing
• perceptual encoding of musical events and
patterns within the context of evoked system ofrelations
• perception of invariance and transformation of musical patterns
• establishment of associative and hierarchical
relations in the building of a mental
representation of musical form.

Event structure processing follows elementary grouping processes. In the perceptual encoding of musical events and patterns, the kinds of values and relations could
include pitch intervals, pitch contours, chord qualities, rhythmic onset intervals, and vectors within points of a timbre space. The encoding process is constrained by
grouping processes, memory limits, and expectations resulting from abstract
knowledge structures. An important aspect of the activation of schemata is the
assignment of relative stability, dominance or salience relations among the events.

Pattern similarity perception is important for musical development, which involves
the abstraction of invariances across transformations. Transformation of musical
patterns is almost universal in musical cultures.

Question 31

Three classes of transformation?

Answer 31

Event structure processing
Perception of invariance and transformation of material• Importance of pattern similarity perception
• Ability to abstract invariances in patterns
across transformation and perception of
relatedness between original and transformed
material
• 3 classes of pattern transformation
▫ direct (linear) operations along a dimension
▫ elaborative modifications
▫ permutations

Question 32

direct (linear) operations along a dimension
▫ elaborative modifications
▫ permutations
involve?

Answer 32

Transpositions: the comparison of the relations among events across
two versions. In (a), the two versions are related by a simple pitch transposition up a
perfect fourth. The perceptual comparison occurs at the level of the representation ofpitch intervals, which are identical in the two versions. One could imagine changes in
tempo of a rhythmic pattern, a change in loudness, and a transposition in timbre
space as examples of this kind of transformation. The second kind of transformation
involves ornamental elaborations of a basic pattern, in which at some level of
abstraction the original pattern still exists. In this case, we say that the elaborated
pattern “reduces to” the original pattern, or that the original pattern can be
“abstracted from” the elaborated pattern. The comparison between the original and
the transformation takes place at a higher hierarchical level. Permutative
transformations involve changing the arrangement, in this case the temporal order is
reversed creating a retrograde of the original. Hardest to perceive. Usually only in twelve tone music, must hear whole chunk, not just second note like others. Changed temporal structure

Question 33

Reduction: hierarchical encoding study

Answer 33

This example from the main theme in the 5th movement of Beethoven’s Pastoral
Symphony (no. 6) shows clearly the notion of reduction and elaboration.
How do listeners encode musical information abstractly? Experimental evidence
indicates that our ability to represent long sequences of events within a hierarchy is
an economical way to encode information.

Diana Deutsch conducted experiments on the role of grouping and reduction in the
mental representation of a musical sequence. Errors in reproduction would indicate
the nature of listeners’ mental representation. The hypothesized hierarchical
reductional structure is shown at the top. The various versions of the structure that
people had to reproduce are shown at the bottom. In (1), there is a natural grouping
due to the repeating ornamental structure that outlines a descending major triad;
participants’ rate of recall was 93.5%. When this grouping is reinforced by rests (introducing temporal discontinuity) at the boundaries between ornamental patterns in (2), the rate of recall was 99%. However, when the grouping by temporal proximity imposed by the silences goes against the reductional structure in (3), it is harder to
encode the patterns in memory (they are not as easily represented as a simpleornamented arpeggio), and listeners have a much more difficult time remembering and reproducing the sequence (rate of recall was 62%). The results show evidence in favour of the hypothesis that the representation of perceived music can be describedin hierarchical form.
Fill interval effect, can structure them, better at recall.

Question 34

Event suturing
Prolongation all reduction
Tree structuring
Dominance of event structures
Hierarchical patterns of tension and relaxation

Answer 34

Lerdahl and Jackendoff propose that the final stage of processing involves the
perception of patterns of musical tension and relaxation. Musical tension or
relaxation established by an event is related to its “prolongational importance” or its
stability, which depends on its place in the tonal hierarchy, the group within which it
appears, its rhythmic value, and its metrical position. Dominant events (e.g.,
important in relation to tonal hierarchy, long duration, strong metric position) would
be “relaxing” in comparison to “tensing” events.

The prolongational reduction attempts to describe graphically these dominance
relations among events. A right branching line indicates a subordinate event followinga more hierarchically important event (e.g., e2 is subordinate to e1). A left branching
line indicates the reverse, (e.g. e3 is subordinate to the subsequent e4). The tree
structure shows the hierarchical network of tension and relaxation. In this way, e1
would be the head event of the major left branch and e4 the head of the major right
branch. e4 has a subordinate relation to e1 at the next level of the tree hierarchy.

Problem is we don’t know to what level we can actually hear all the other levels also. Music theory analysis take birds eye view

Question 35

This experiment on simple binary folk songs was conducted by Thomas Stoffer. He
played the songs designed to have a very clear hierarchical structure shown in Fig. 1. On positions 8 through 24 he played a click and asked listeners to locate the click in
the score. In reality, they were evenly distributed across all positions, but as Fig. 2
shows, the places where people heard them correspond to positions at structural
boundaries as indicated by the hierarchical structure (phrase boundary at 16 and
subphrase boundaries at 8 and 24). The stronger the structural boundary, the more
often people heard the click there. Indeed as Fig. 3 shows, there was a tendency to
displace the clicks toward the nearest boundary (i.e., the displacement is positive
before the main middle boundary and negative just after it). This suggests that thereis some correspondence between the theorized formal description of phrase
structure and the representational structure of the perceived music. In addition,
listeners’ perception of the temporal position of extra-musical events is somehow
affected by their mental structuring of the music itself.
Another experiment?

Answer 35

Emmanuel Bigand and Richard Parncutt tested the idea that prolongational reduction would give rise to variations in musical tension by playing a sequence of 31 chords
that had a theorized hierarchical structure indicated by the reduction tree. The
progression involved local intra-key hierarchies as well as modulations to other keys
(first to F major and then to B Major) creating an inter-key hierarchy. Listeners heard
chords 1-2 and had to estimate the degree of perceived musical tension on a scale of
1 to 10. Then they heard chords 1-3, 1-4, etc. and each time estimated the tension.
They predicted that the higher in the tree a change takes place, the stronger should
be the sense of a musical boundary, and the more people should indicate low tension at that place. The graph at the bottom shows the mean tension ratings for chords 2-
31 for musicans and nonmusicians. Note that the tension profile follows higher levels of the reduction tree (lower tension ratings correspond to larger branches of the tree) and that there is very little difference between musicians and nonmusicians.

Question 36

Problem with tension ratings

Answer 36

Only works for chunk, as modualtion cause problems, recency effect.
Compromises.

Question 37

Memory compromise for event structure processing

Answer 37

Bigand (1993) proposed a dynamic model of how grouping, metric and time-span
reduction structures are assembled within a sliding window representing a buffer of
working memory. This figure provides a useful overview of the lecture content on
event structure processing.

In A, the listener experiences musical events in successive zones, symbolized by a
sliding window. In B, for each zone, the listener extracts the metric structure of
events (represented by black dots) and one or several levels of grouping structure
(represented by parentheses). In C, the musical information contained in each zone isthen organized and represented hierarchically.

One big question is the extent to which the tree relations extend very far in time, i.e., how much of a musical form can we really represent mentally.

First five seconds, most important taken, rest dropped. This goes on, other things grouped. .