PSYC56 Music Cognition
Lecture 8 Musical Expectancy
Time, Rhythm, and Meter
The Formation of Musical Expectancies
Meyer talked about implicative patterns or patterns that would imply a continuation or generating
expectancy based on Gestalt notions. There is a structural leap; we expect that to be filled in, as a
notion of closure or completion.
There can also be a pattern increasing, we would expect a tone that would continue that pattern.
According to Meyer, on the basis of patterns, we could have expectations for the notes that are going
to come next.
Schmuckler was interested in what drives our expectations to occur at different times. He took eight
measures of a complex musical passage. Looking over this passage, based on melodic features, was
able to identify 10 different propositions.
Listeners would hear the passage and said it would go one more note, and asked listeners to rate how
that element fit in with expectations that were going to fit in at that point. In one they heard melody
line, the harmonic accompaniment, or both.
At PP2 there is pattern that is going up, so we may expect continuation based on the linear pattern.
In context, heard a melody, then trial 1 heard a note after the melody, and was asked to rate how well
that note fit in.
The note F got the highest rating that was the note listeners predicted was going to occur. Listeners
gave a very high rating to the note F.
More generally, in Probe Position 2, listeners rating for their expectations were very much driven by ,
tested Meyer’s notions of gap fills and linear patterns, second factor that was tested was the idea of
tonality, this piece is in a particular key, and maybe what listeners expect are tones that fit in well
with it, a third factor looked at was melodic contour, how well was general pattern of the contour
The open squares are the strength of the prediction based on ideas from Meyers, the open circles are
the predictions based on tonality, how well the tonal structure drives it, the closed triangles are
predictions based on contour and this is across 10 different propositions.
Generally found expectations were contributed by tonality and melodic process contributed
significantly. Contour never predicted listeners’ predictions.
Note both open factors predicted listeners’ expectations and they had a relationship, when melodic
process got stronger, tonality decreased, and when tonality increased, melodic process got weaker.
Melodic process is the name Meyer gave to Gestalt patterns.
We can model and look at listeners expectations to what is occurring through a musical passage and
we can make predictions based on melodic processes and tonality. Listeners expectations are
predictable and we can develop models for what we think is going to occur next.
The Implication-Realization Model
Narmour defined 5 different rules for patterns of Meyer, and these rules drive our expectations.
If there is a small interval between two notes, you are going to expect another small interval in that
same direction. But if you have a large reversal, you are going to expect the next interval to come
down- registral direction.
We expect small intervals not large intervals- intervallic difference. Narmour took these rules and applied them to passages of music. He used this to provide theoretical
analysis and to show where our expectations would be strongest in a piece. This model actually
suggests that this is the core of one’s expectations and we would be perceiving intervals in keeping
with these types of expectations.
Is there any evidence, and do we see listeners expect intervals to occur based on these principles?
Schellenberg took Narmour’s ideas and used them so that he can drive predictions. He quantified
Narmour’s rules and quantified them. *
He played listeners 2 notes then played a 3 note and asked how did this fit their expectations and he
did it for all the implicative intervals, and then looked at listeners ratings and try to predict what
their expectations would be based on principles.
He found pretty convincing evidence that these principles seem to work. He found evidence for these
factors but not all of them.
Simplification of Narmour’s Implication-Realization Model
Schellenberg was able to simplify his model to two factors, the notion that is you have a implicative
interval, you have some expectations for it to be the same size interval in the opposite direction. This
was a combination of proximity, intervallic difference, and closure on slide 11 gave the structures on
slide 12. Tonality was also important for driving expectations.
Also found principles occurred cross-culture, and drove expectancies in other cultures. This is
interesting because these factors are presumably based on low level Gestalt principles. Student asked
if musical expectancy is innate? To prove that something is innate, you have to improve there is no
experience that impacted this, and you have to test at the moment of birth which is hard because
there is 9 months of experience in utero. If principles are common across culture, then you wouldn’t
argue innate as you would say there may be a genetic basis for these things, that doesn’t mean
experience isn’t important in bringing them out.
In previous study, if you look at staff B in slide 6, we can derive information for what chord is going to
come next, and we can use Piston’s table of usual root progressions.
You see different scores of people’s expectancy, if you take the rating the chords that Piston’s said
would follow and subtract what he said may follow, and this is driving listener’s expectations, the
cords that often follow should have higher ratings than sometimes follow which should have higher
rating than seldom follow. This works for the top often follows and sometimes follows. The
sometimes don’t get higher than seldom, and seldom doesn’t get higher ratings than never. But at
what chords are really expected vs. less expected, we can predict listeners ratings based on Piston’s
Combined (melodic and harmonic) expectancies
Other than one place (at 9 where harmonic did not contribute significantly) generally this is saying
listeners expectancies of combination melodic and harmonic were added combination of melody plus
This additive model was tested against other interactive models, and the interactive model never
predicted better than additive model.
Information given on slide 6 Schmuckler
o Propositions refer to the different points in time. PP1, heard the context up to that point, and
a bunch of continuations, then PP2, and a bunch of continuations, then PP3, then a bunch of continuations, until they got to PP10 they heard the whole 8 measures and rated the final
o One finding that was cool with the 3 conditions, melody only, harmony only, then combined
melody and harmony. By the time they enter the third condition, the listeners have heard the
entire piece, one of the things you can ask is what role does prior experience have in
modifying their expectations.
o The study did counterbalancing, where half heard full context, then moldy and harmony,
then other half heard melody and harmony than full context. Found prior experience had no
impact. Knowing what was going to occur did not change listener’s expectations. There are
couple of places where there are surprises have such as PP1 that lower note, so that note
raised listener’s expectations.
o At PP6 listeners expected B flat not A flat, and when they heard the piece, expectancy of A
flat but B flat was still the highest expectancy.
o If expectancy is so important, why would we ever want to hear a piece of music a second
time? This passage we have musical surprise, something unexpected occurs. We cannot be
surprised by something we know is going to occur. How can we have musical surprise from
music we have heard before? If expectancy is driven by low level factors, doesn’t matter that
we have heard it before, they are still going to operate.
The Harmonic Priming Paradigm
We have a notion of some response to the question of what questions drive our expectations.
We can also talk about the consequences of expectancy formation. What happens to us of our
expecting certain events, how we perceive the passage and remember the passage? Does expectancy
generation have an impact on how we understand the music?
One thing we can propose is maybe there is a processing advantage to processing events such that an
expected event is perceived and processed quicker than unexpected event.
Stoeckig wanted to look at notion of impact on processing. If you have a particular concept or object,
items that are highly semantically related to that are easier to respond to, it is like that initial item
primes you to respond to another item. Semantic priming such as asking if this is a word after you
prime with another word such as table, people will be quicker to say chair is a word than apple is a
word based on table prime.
Listeners heard a first chord, than a second chord. The general idea is tell me whether the second
chord is in tune or out of tune. The first volume is in tune and the second is out of tune.
In unrelated context, the first was in tune and the second was out of tune.
Stoeckig found a priming response. When in tune targets, the unrelated chords are processed slower
than related chords. Just to say in tune, you are much quicker when second chord is harmonically
related to the first chord. You see the reverse for out of tune, it takes more time for you to say that
the related chord is out of tune than the unrelated chord. You get the same effect in error rates.
This is saying that expectancy, the idea that first chord sets up an expectation for what you are going
to say next has an impact on how quickly and accurately you can perceive that upcoming chord.
The harmonic priming paradigm became fertile ground for researchers and they explored a huge
wealth of questions. This research has been extremely influential.
Harmonic context and phoneme monitoring
Argued that the harmonic structure is affecting processing of a number of musical passages. The
generation of expectation leads us to processing advantages for any number of musical dimensions.
There is no reason