-
Musical Improvisation and the Brain:
A Cross-Cultural EEG Study of Jazz and
Hindustani Musicians
Honors Thesis Submitted by
Marion Wellington
in partial fulfilment of the
Sc.B. in Music Cognition
Brown University
April 15th, 2016
Prepared under the Direction of
Dr. Michael Worden, Advisor
Dr. Monica Linden, Reader
Dr. Dana Gooley, Reader
-
-2-
Table of Contents
Acknowledgments...........................................................................................................................
4
Abstract
...........................................................................................................................................
5
1: Introduction
.................................................................................................................................
6
1.1: Review of Improvisation in Music Cognition
......................................................................
6
1.1a: Role of Improvisation in Music Therapy
......................................................................
15
1.2: Lack of Cross-Cultural Studies in Music Cognition and
Real-World Applications in Music
Therapy......................................................................................................................................
17
1.3: Discussion of Jazz and Hindustani Musicology
................................................................
19
1.3a: Jazz theory
....................................................................................................................
19
1.3b: Hindustani theory
.........................................................................................................
20
1.3c: Comparison of jazz and Hindustani improvisation
...................................................... 23
1.3d Comparison of the pieces: On the Sunny Side of the Street
and the Raga Yaman ....... 28
1.4: Objectives
...........................................................................................................................
29
1.5: Hypotheses
.........................................................................................................................
29
2: Methods
....................................................................................................................................
31
2.1: Participants
.........................................................................................................................
31
2.2: Materials
.............................................................................................................................
33
2.3: Protocol
..............................................................................................................................
34
3: Results
.......................................................................................................................................
36
3.1: Preparation and selection of the data
.................................................................................
36
3.2: Descriptive Analysis
..........................................................................................................
38
3.2a: Spectrograms
................................................................................................................
38
3.2b: Averaged Alpha
............................................................................................................
40
4: Discussion
.................................................................................................................................
42
4.1: Discussion of the
results.....................................................................................................
42
4.2: Limitations of this
study.....................................................................................................
46
4.3: Suggestions for future studies
............................................................................................
49
5: Conclusion
................................................................................................................................
51
6: References
.................................................................................................................................
52
7: Appendix A: Musical Terms
.....................................................................................................
57
-
-3-
Western......................................................................................................................................
57
Indian
.........................................................................................................................................
61
Other
..........................................................................................................................................
63
8: Appendix B: see folder
.............................................................................................................
64
-
-4-
Acknowledgments This study was only made possible by an enormous
collective of people, for whom I’m very grateful. To Michael
Worden, for accepting the half-baked study I brought forth and
helping me make it into something of which I’m proud. To Monica
Linden, for facilitating my interest in this field since day one
and helping me navigate my... singular undergraduate career since.
To Dana Gooley, for helping me bridge the gap between science and
humanities with curiosity and enthusiasm. To Carlos Aizenman, for
taking on my Independent Concentration and giving me as many
resources as you could muster. To Chuck Royce, for not only giving
undergraduates the opportunity to delve into their passions and
create unorthodox projects, but also for giving us the freedom to
fail. To Kerrissa Heffernan, for telling me I’m “weird, and that’s
a good thing.” To David Sheinberg, for lending me your EEG system
on our first meeting. To The Moore Lab, for letting me pipette
electrode gel onto the head of one of your lab members. To
Stephanie Jones, for giving me both advice and reassurance. To
Peggy Chang, for asking critical questions while giving absolute
support. To Peter Bussigel, for lending me technical equipment and
giving me life equipment. To Tom, for encouraging me to go for it.
To Shantala Hegde at the National Institute of Mental Health and
Neuro Sciences and Stephen McAdams at Music Perception and
Cognition Lab at McGill University, for responding to my
out-the-blue emails not just with a yes, but with continuous warmth
and guidance throughout my summer and beyond. To Deepak Ullal, for
attempting to fix the EEG headset not once, but twice, and sharing
your joy and skill with the santoor. To Eric Lewis, for hashing out
experimental design with me and showing me a prime example of
Montreal’s free jazz. To Hélène Martel, for supplying me with
participants in an unknown city. To all of my participants, for
bringing patience and excitement to each trial. To Colette, Mahoro,
Robert, Kalie, Dhanya, Nigesh, Shreena, Anushka, Neil, and Ria, for
helping me add new meaning to my definition of home. To Bryn and
Jamie, for not only making the GISP that started it all, but also
inspiring me to further myself as a scientist and a musician. To
Morgan, for being my music cognition rock in a campus frustratingly
fixed on the visual system. To my friends, for encouraging me to do
what makes me happy and reminding me to take care of myself.
Finally, to my parents, for pushing me to do well, pushing me to do
good, and supporting me in whatever that means to me. I love you.
Thank you all so much for helping me get to the place I am today. I
am overwhelmed with gratitude and appreciation.
-
-5-
Abstract Musical improvisation is a highly complex creative
process for which there is growing
interest in the field of music cognition. It has been studied
from a variety of perspectives, from
classical string players in front of an audience to jazz
pianists trading solos to freestyle rap
artists. These studies have pointed to an expertise-related
difference in brain activity, supporting
Pressing’s model of practice-contingent automation of
lower-level processes. However, the
current literature on improvisation is quite Western-centric,
with no present studies of
improvisation outside the Western tonal system. I attempt to
reconcile that gap by conducting an
electroencephalography (EEG) study of improvisation in jazz and
Hindustani musicians to
examine the differences between performing composed music and
improvising and to compare
these differences across cultures. I focused on the alpha band
frequencies (7.5-12.5Hz) in the
frontal and posterior regions, as high alpha power has been
found to correspond with divergent
thinking tasks such as improvisation. Though the Hindustani
participants had low alpha power
during the composed condition, they consistently had higher
alpha power than the jazz
participants during the improvised condition, both in the
frontal and posterior regions. Both of
these findings may be indicative of Hindustani music tradition,
both in its melodic complexities
and demanding practice time; the Hindustani composed content may
have been inherently more
difficult than the jazz content and overall the Hindustani
participants practice more frequently
and at greater length than do the jazz participants. This study,
though not comprehensive, acts as
a pilot study for future inquiry into this field of
cross-cultural musical improvisation.
-
-6-
1: Introduction Improvisation is the act of spontaneous musical
composition. It is a form of artistic
creation that is physically and temporally restricted; it
incorporates elements of composition and
performance together in real time. It follows that improvising
requires extensive and expansive
cognitive function. As the fields of music cognition and the
neuroscience of creativity are
becoming increasingly popular, the neuroscientific study of
musical improvisation is also
garnering interest in the scientific community. However,
existing studies have been quite
Western-centric. My thesis seeks to address this disparity by
approaching the study of
improvisation from a cross-cultural lens, by analyzing
electroencephalography (EEG) recordings
from both jazz and Hindustani musicians as they improvise and
perform previously composed
music.
1.1: Review of Improvisation in Music Cognition
Because improvisation is a creative process, the way in which it
is conducted in an
experimental setting can vary widely. Most of the current
literature uses pianists, from either
classical or jazz backgrounds, and compares brain activity
recorded with fMRI or EEG between
playing composed music and improvising with varying conditions.
Researchers have asked a
variety of questions in their studies, but they all investigate
the neural substrates of
improvisation. Table 1 shows an overview of the literature
reviewed.
-
-7-
Table 1: Abbreviated summ
ary of musical im
provisation literature review
-
-8-
Though the reviewed studies highlight a diverse range of brain
regions, there were key
regions that were consistently implicated, either by their
significant activations or deactivations.
The pre‐supplementary motor area, which is implicated in the
temporal control of motor
sequences, was found to be consistently activated during
improvisation (Bengtsson et al., 2007;
Liu et al., 2012; de Manzano and Ullén, 2012; Pinho et al.,
2014; Donnay et al., 2014). The left
inferior frontal gyrus, or IFG, is associated with retrieval of
long term memory, especially in
relation to language; it too was active in multiple studies
(Limb and Braun, 2008; Berkowitz and
Ansari, 2010; Liu et al., 2012; de Manzano and Ullén, 2012;
Donnay et al., 2014). The IFG is
often activated in syntax retrieval in language, implying that
musical improvisation could be
utilizing syntactical musical knowledge from long term memory to
generate novel musical
sequences. The dorsal premotor cortex, which projects directly
to the spinal cord and is involved
in volitional control of movement, also showed activation
(Bengtsson et al,. 2007; Berkowitz and
Ansari, 2010; de Manzano and Ullén, 2012; Liu et al., 2012).
Two other regions, the dorsolateral prefrontal cortex, or DLPFC,
and the medial
prefrontal cortex, or MPFC, are contested in their roles in
improvisation. The DLPFC is the
endpoint of the dorsal pathway, which ‘tells’ other brain
regions how to interact with stimuli. It
is a major site of executive function, working memory, active
decision making, and cognitive
flexibility (Limb and Braun, 2008). In contrast, the MPFC is
associated with spontaneous
thought processes, such as mind‐wandering (Limb and Braun,
2008). The DLPFC has been
characterized as a location of executive function that interacts
with external attention and stimuli,
while the MPFC is thought to be a location of self‐generated
stimulus‐independent cognition. In
some studies, there was activation in the MPFC and deactivation
in the DLPFC, implying that
improvisation might be a largely internal, subconscious
mechanism (Limb and Braun, 2008; Liu
-
-9-
et al., 2012; Pinho et al., 2014). It supports the concept of
suppression of executive control
systems and the activation of default mode regions in
improvisation. As Limb and Braun state,
[the activation of the MPFC] may reflect a combination of
psychological
processes required for spontaneous improvisation, in which
internally motivated,
stimulus-independent behaviours unfold in the absence of central
processes that
typically mediate self-monitoring and conscious volitional
control of ongoing
performance (2008: 4-5)
In other studies, the DLPFC showed activation, while the MPFC
did not (Bengtsson et al., 2007;
de Manzano and Ullén, 2012; Donnay et al., 2014). Those studies
argued that sensory input
integration and selective retrieval are crucial to the act of
improvisation.
So, how can these competing arguments be reconciled? The answer
may be in line with
Jeff Pressing’s expertise-contingent model of improvisation
(1988). Pressing postulates that
musical improvisation is an acquired skill through extensive
training. Improvisation is
demanding of many mental faculties simultaneously, such as
sensory encoding, motor control,
and memory retrieval. Pressing believes that practice automates
such lower-level processes,
allowing the performer to focus their faculties on higher-level
processes, such as musical idea
generation and evaluation.
In the studies in which the DLPFC was active, the musicians were
either “trading fours”
(where each musician takes a solo every four measures) with
another musician or they were
classically trained musicians asked to improvise. The former
condition requires the integration of
sensory input from their fellow musician, and the latter
involves musicians with expertise in a
related but not exactly similar field. The studies that involved
activation in the MPFC (and
-
-10-
deactivation of the DLPFC) used jazz musicians at a least a
semi‐professional level. Classical
musicians are rarely asked to improvise, while improvisation is
an integral element of jazz. So,
though the classical musicians had extensive musical expertise,
their lower‐level processes may
not be as automated as the musicians that improvise regularly.
Conversely, those with
improvisational expertise may have those processes more
automated (as Pressing predicted) and
can thus allocate their faculties to higher‐level processes.
There are other ways in which expert improvisers differ from
those without expertise.
Pinho and her colleagues studied participants with varying
experience in both jazz and classical,
stating that
...greater functional connectivity of the frontal brain regions
seen in the most
experienced participants may reflect a more efficient
integration of
representations of musical structures at different levels of
abstraction. A higher
functional connectivity of the seed regions was observed with
premotor regions
and parietal and prefrontal association cortex, as well as with
primary
sensorimotor cortex and the cerebellum, suggesting that the
training-related
functional reorganizations may affect both cognitive and
sensorimotor aspects of
improvisation. (Pinho et al., 2014: 6161)
In the studies that used EEG, all three suggested a more
widespread connectivity in the
frontal and sensory regions, supporting Pinho’s postulation.
This higher functional connectivity,
along with the activation of the MPFC and the deactivation of
the DLPFC in expert improvisers,
all points to a top-down process in which attention is directed
less at lower-level processes such
-
-11-
as motor sequences and planning, but instead to more upper-level
ideas such as overall form and
divergent thinking. Limb and Braun state that
It has also been suggested that deactivation of the lateral
prefrontal regions
represents the primary physiologic change responsible for
altered states of
consciousness such as hypnosis, meditation or even daydreaming.
This is
interesting in that jazz improvisation, as well as many other
types of creative
activity, have been proposed to take place in an analogously
altered state of mind.
Moreover, a comparable dissociated pattern of activity in
prefrontal regions has
been reported to occur during REM sleep, a provocative finding
when one
considers that dreaming is exemplified by a sense of defocused
attention, an
abundance of unplanned, irrational associations and apparent
loss of volitional
control, features that may be associated with creative activity
during wakefulness
as well. (2008: 5)
Alpha waves are oscillations in the frequency band of 7-13Hz
that are usually associated
with a relaxed state of mind, similar to what is described
above. Indeed, other EEG studies of
creativity (though not specifically improvisation) have pointed
to alpha waves as being widely
prevalent during creative processes. A robust review on the
neuroimaging of creativity done by
Arden et al. (2010) argues that the current literature is too
disparate to make any general
conclusions. However, it did state that studies that used the
Alternate Uses Test (or a variant on
such) all showed high levels of alpha synchronization. The AUT
is a test done to measure
creative ideation; the classic example is to give the subject a
common object (e.g. a brick) and
ask them to list as many possible uses for the object as they
can think of (e.g. as a building for a
-
-12-
mini King Kong to topple in a reenactment). It demonstrates
divergent thinking, which is the
generation of many possible solutions to an open situation. As
Fink and Benedek state, “It
[divergent thinking] is conceptualized as a cognitive process
involving both the retrieval of
existing knowledge from memory and the combination of various
aspects of existing knowledge
into novel ideas” (2012). Musical improvisation can easily be
classified as a type of divergent
thinking; as a solo progresses, the musician generates multiple
concepts of how the phrase can
start, follow through, and end, given the musical framework in
which they are operating.
Fink and Benedek (2012) addressed Arden et al. (2010) in their
own review of creativity
literature. They argued that their summary was inconclusive
because they were not focusing on
any particular aspect of creativity. Fink and Benedek reviewed
literature that related specifically
to divergent thinking. Consequently, they found an overall trend
of more alpha power in
divergent tasks than rest or convergent tasks. The more creative
the task, the higher the alpha
levels present. Fink and Benedek also found that frontal alpha
synchronization correlates with
top-down processing, a trait of creative processes such as
musical improvisation. To clarify the
relationship of alpha waves to divergent thinking, Jauk et al.
recorded EEG of participants who
did convergent and divergent tasks, and found that while
convergent thinking resulted in a strong
task-related desynchronization of alpha activity, divergent
thinking resulted in synchronization
of alpha activity, especially in frontal cortical sites
(2012).
The prevalence of alpha synchronization (a frequency band most
associated with
relaxation) and the deactivation of lateral prefrontal regions,
specifically the DLPFC (which has
been associated with daydreaming, defocused attention, and other
altered states of consciousness
by Limb and Braun, 2008) point to the presence of “flow” in
improvisation and other creative
processes. Csikszentmihalyi defines flow as “an almost
automatic, effortless, yet highly focused
-
-13-
state of consciousness” (1996) in which a task is performed to
the best of an individual’s ability.
Though the term stems from positive psychology and is referenced
in a variety of fields, many
neuroscientists have become intrigued with the phenomenon, its
neural underpinnings, and how
it relates to creativity. Flow has been cited to occur in many
different situations, such as sports or
videogames, but it is most researched as it relates to creative
processes, such as writing, theater,
and improvisation, because it is thought that flow and
creativity are inextricably linked. Dietrich
identifies free-jazz improvisation as one of the more common
situations in which someone can
experience flow:
In free-jazz improvisation, the musician arranges units into a
flowing string.
Because the string progresses by each unit triggering the next,
the application
becomes part of the procedure. The overall product can be novel
(indeed, if the
string is long enough it must be novel due to the complexity of
the musical
system). The full string can even be multi-dimensional, but each
individual step is
not. It is the number of distinct reflexive loops as well as
their level of
automatization that determine the quality of the flow
experience. It should be
noted that such increased implicit expertise does not
necessarily lead to the skills
representation in the explicit system. (2004: 756)
Dietrich believes that flow comes out of the interplay between
explicit and implicit
learning, and that “optimal performance involving a real-time
sensory-motor integration task is
associated with maximal implicitness of the task’s execution.
Given that the explicit system is
subserved by prefrontal regions, it follows from this proposal
that a flow experience must occur
during a state of transient hypofrontality that can bring about
the inhibition of the explicit
-
-14-
system” (2004). This concept of hypofrontality intersects with
the evidence of expertise-related
deactivation of the DLPFC, and the corresponding frontal alpha
synchronizations.
Though many music cognitivists conduct pure research for the
sake of understanding
music in the brain, the field can also be used in a direct
application to health: music therapy.
-
-15-
1.1a: Role of Improvisation in Music Therapy Music therapy is a
burgeoning field in health care, one that is beginning to gain
more
credibility as a substantive way to assist medical treatment.
Music therapy has been successful in
a variety of health issues, from alleviating mental illness to
physical and neuro-rehabilitation in
stroke victims. While there are many forms of music therapy
(listening, playing, and creating
music to name a few), improvisation is being used increasingly
in a variety of treatments to
engage its complex cognitive properties. In a robust review of
the current literature regarding
improvisation and health, MacDonald and Wilson note
“Improvisation in music therapy is seen
to have specific benefits for particular populations including
the amelioration of neurological
damage, improvements in mental health conditions, reductions in
stress and anxiety, and
improved communication and joint attention behaviours in
children with autistic spectrum
disorders” (2014). They attribute the therapeutic effects to
multiple aspects of the practice. The
mental faculties that improvisation demands, whether or not one
is a seasoned musician, are
considerably large. This puts the patient in a state of focus
through a creative process, in a way
allowing them a reprieve from otherwise taxing states of mind.
Second, improvisation creates a
venue for the patient to express their emotions in a non-verbal,
non-explicit way; patients who
have difficulties articulating themselves can release those
often repressed feelings in a productive
manner (Wigram, 2004; Smeijsters and van den Hurk, 1999;
O’Callaghan, 2004; Volkman,
1993). Lastly, the non-verbal social interaction that paired or
group improvisation provides has
shown significant improvements in populations who otherwise have
trouble interacting, such as
patients on the autism spectrum (Geretsegger et al., 2012;
Simpson and Keen, 2011).
While most studies of music therapy and its effect deal with
behavioral results only,
others have recorded biological and neurological data to
corroborate the behavioral findings. In
studies of patients with neurological damage, usually due to
stroke, scientists have found
-
-16-
remarkable physical improvements in patients who improvise,
usually due to the rhythmic
entrainment of the musical interaction (Aigen, 2009; Tomaino,
2013). Fachner (2012) recorded
EEG data of 62 patients with depression and comorbid anxiety.
Half of the patients received
improvisational music therapy in addition to their standard care
while the other half served as
controls (receiving standard care alone). By comparing
recordings done at both the inception of
the treatment and three months in, they found that
improvisational music therapy not only
significantly reduced depression and anxiety symptoms, but also
elicited long-term absolute
alpha power increases in the left fronto-temporal lobe and theta
power increases in the left
fronto-central and right temporoparietal lobes.
-
-17-
1.2: Lack of Cross-Cultural Studies in Music Cognition and
Real-World Applications in Music Therapy
Clearly, the inquiry into the improvisational brain is growing.
However, all of these
studies (not unlike the rest of music cognition studies) are
Western-centric; they focus on
improvisation purely in the context of the Western tonal system,
whether it be jazz, classical, or
more recently, hip-hop. But there are many genres of music all
around the world that have strong
themes of improvisation. Middle Eastern folk, Indian classical,
Javanese Gamelan drumming,
Native American folk, West African Mande, and more incorporate
improvisation into their
compositions and performances. So why has the focus been
exclusively on Western jazz and
Western classical? One can argue that we have easier access to
these participants, and keeping
the material studied constant allows for more congruence between
studies. However, that narrow
parameter precludes the possibility of discovering whether the
neural phenomena studied are
inherently human or just inherently Western.
There is a small body of work in music cognition that addresses
this disparity by
comparing aspects of Western music to its non-Western
counterparts. A study on musical phrase
processing in German subjects while listening to Western and
Chinese excerpts showed that
there was a much stronger sensori-motor network engaged while
listening to culturally familiar
music versus unfamiliar (Nan et al., 2008). This implies that
our musical entrainment is not all
innate, but is partially due to the prevalence of the tonal
system in our everyday lives. A study in
cross-cultural music comprehension in which Western musicians
and nonmusicians listened to
Western and Chinese musical excerpts showed that although the
neural correlates while listening
did not significantly differ across the two genres, the
subjects’ recall performance was much
poorer for the non-familiar genre (Morrison et al., 2003).
Lastly, a study was recently conducted
in which the emotional response of both Westerners and Pygmies
were recorded as they were
-
-18-
exposed to genres from each culture (Egermann et al., 2015).
They showed that while the
perception of the music’s emotional value varied across
populations, physiological arousal due to
low-level acoustic properties was relatively constant across
cultures. This model of comparing a
specific component of music within a Western paradigm and a
non-Western paradigm is quite
effective for discerning what is culturally derived and what is
naturally innate. Unfortunately,
such a modest body of work fails to fully address the
nature-nurture debate.
This investigation of improvisation cross-culturally is intended
to add to this
conversation, as currently there exists no such study in the
literature. The study of improvisation
from a cross-cultural lens is important because it helps remove
a cultural bias from the results. If
there are commonalities between the two groups, this may imply
innate abilities or cognitive
processes not molded by their environment. If there are
dissimilar properties, hypotheses can be
drawn about how their respective cultures influence the data.
Further, these findings could
potentially make improvisation a more accessible health tool to
other cultures, possibly globally.
It has been shown that improvisation can be used as a powerful
tool in music therapy and
health. However, as stated previously, the vast majority of
these studies have been conducted in
the Western sphere with Western patients in mind. If the
practices of improvisation between
Western and Indian cultures (however they may differ
theoretically) have similar neural
substrates, we can suggest that improvisation can be utilized as
a tool for health in India as well
as the West.
This study focuses on improvisation within two musical groups,
Western jazz and Indian
classical Hindustani music. I have collected EEG data from
vocalists in both traditions as they
perform previously composed music and improvise within their
common musical parameters.
-
-19-
1.3: Discussion of Jazz and Hindustani Musicology
Why choose jazz and Hindustani as the two forms of music to
compare? While the genres
come from much different origins, they have elements of both
composition and improvisation in
their performances. The following section examines the theory
and practice of improvisation in
both genres, and how they may be compared. See Appendix A for a
glossary of musical terms.
1.3a: Jazz theory Jazz is difficult to define, as it had evolved
both linearly over more than a century and
laterally over a multitude of genres and subgenres. Its roots
are in New Orleans in the early
1910s, when Black Americans combined ragtime, brass bands, slave
rhythms and melodies,
blues, African rhythms, and more, using a polyphonic form of
improvisation where musicians
would weave their sounds in and out of each other’s in a state
of harmonious cacophony. Louis
Armstrong is credited with introducing solo improvisation to the
greater jazz world, in which a
single musician would take the helm as the others would quiet
into the background and would
then take turns. Over the course of the 20th century, different
flavors of jazz took center stage,
from stride to big band to swing to bebop to hardbop to modal
jazz to free jazz, with subgenres
spinning off and influencing other genres, such as Cu-bop, RnB,
soul, funk, smooth jazz, and
other jazz fusions.
Though jazz has been through many epochs and styles, there are
some overarching
themes that remain relevant to the larger genre. For the
purposes of this paper, I will focus on the
melodic and harmonic content of swing, as that is the area of
jazz used in this experiment. Swing
is characterized as a type of performance style, popularized by
big bands in the 1930s, that has
an idiosyncratic bounce in which the eighth notes are performed
such that the downbeats and
upbeats have ⅔ and ⅓ of the beat respectively. A good example of
a tune that switches between
https://docs.google.com/a/brown.edu/document/d/1h21bz9GcU7dsJhUfnyJ5fcE018FPRJ3SZ8ZFe3gyjpo/edit?usp=sharing
-
-20-
swing and “straight” playing is “Caravan” by Duke Ellington, in
which the chorus starts with a
straight feel and then jumps into swing feel. The repertoire
used in swing, and in many other
iterations of jazz, are called standards, having come from jazz
composers, Broadway show tunes,
Hollywood musicals, and other sources. These standards (and
other jazz tunes) have a melody
(called a head, or the chorus) and harmony (called a chord
progression, or changes) that can be
altered by performers often to great lengths. A hallmark of jazz
harmony is the use of seventh
chords, rather than triads, as the building blocks for a
progression. Additionally, another pillar of
jazz harmony is the authentic cadence (V-I) preceded with a
predominant ii chord, ii-V-I.
Tensions, unusual scales, and modes are all devices used in
composition, reharmonization of
standards, and improvisation.
The chord changes are usually supplied continuously throughout
the piece by a
polyphonic instrument, like a piano or guitar, with drums
supporting the rhythm and the bass
supporting the harmony. A vocalist, a horn, or any other
monophonic instrument usually
performs the head, though any other instrument (save the drums)
can supply that. Every musician
in the combo, or big band, is able to improvise on the piece
(including the drummer). A melodic
improviser needs to know the restraints of a tune: the meter,
form, and harmony dictate the
directions the musician can go in their improvisation.
1.3b: Hindustani theory Hindustani music is as old as jazz is
new, as static as jazz is dynamic. Hindustani music
in its pure form is relatively untouched by modern music styles
and forms. This is evident in its
form, the overall structure of a piece. Jazz pieces follow the
form found in Western classical and
pop music in which they have very distinct sections that are
repeated later on. Hindustani music
follows a more linear form. There are distinct movements, but
they are not usually repeated.
-
-21-
Likewise, the improvisation/composition aspect is much less
dichotomized in Hindustani music
than in jazz music, as will be discussed further on. These
differences are most likely due to both
the geographic and temporal distances between the genres.
Hindustani music is passed down from guru to shishya without
written notation. The
musical content of the Hindustani tradition is remarkably
complex and requires decades of
training to master. Vocalists usually start in their early
adolescence and will practice a few
selections for many hours every day. While Western and Bollywood
music have influenced the
mainstream Indian music heavily, the classical genres have
stayed relatively constant, reflecting
the values of tradition and spirituality that many of the
musicians possess.
Unlike jazz and other Western musics, which have rhythm, melody,
and harmony as their
main building blocks, Hindustani music does not regard harmony
highly. Instead, the melody
and rhythm, the raga and the tala, are what drive the
performance. In place of harmonic changes
is a drone, usually from an instrument called the tanpura, which
resonates at the shadja and
pancham swaras of the sargam (the tonic and dominant notes of
the scale in Western
terminology). The tala is usually driven by the tabla, a
percussion instrument that can be
manipulated to have a bent pitch. A tala is a repeating,
rhythmic phrase that cycles at a variety of
beats with specific subdivisions. It does not have a fixed
tempo, and can speed up or slow down
at the will of the performers during a piece.
There are many instruments that can play the melody, such as the
sitar and santoor, but
the voice is the primary melodic instrument of Hindustani music.
The raga (also known as a
raag) is the melodic backbone of a performance. Its closest
Western analogue is a mode, though
in Hindustani music the raga is more than a scale; it is a
color, a feeling, a passion. As Ravi
-
-22-
Shankar, the prolific sitar player attributed with popularizing
Indian classical music in the West,
puts it:
A raga is a scientific, precise, subtle and aesthetic melodic
form with its own
peculiar ascending and descending movement consisting of either
a full seven
note octave, or a series of six or five notes (or a combination
of any of these) in a
rising or falling structure called the Arohana and Avarohana. It
is the subtle
difference in the order of notes, an omission of a dissonant
note, an emphasis on a
particular note, the slide from one note to another, and the use
of microtones
together with other subtleties that demarcate one raga from the
other.
Every raga has unique pitch content and intricate pitch
relationships; the performance
and exploration of a raga can be up to an hour long. There are
thousands of ragas; most vocalists
learn the most common ragas and will choose a couple to work on
for many years. It is not just
the pitch content, but the way in which certain notes are
prioritized and approached that
comprises a raga. The vaadi and the samvaadi are the two most
important notes, aside from the
home note, and the way they are approached can distinguish two
ragas with the same pitch
content. The scales themselves, the sargam, are quite comparable
to Western scales. The
shortened names of each of the basic seven swara are used in the
same way solfege is used in
Western music: Sa Re Ga Ma Pa Dha Ni (Sa), and the natural scale
is the same as the Western
Ionian, or major scale. The swara R, G, D, and N can be either
natural or flat, the M can be
either natural or sharp, and S and P can only be natural,
totaling in the same pitch content as the
Western chromatic scale. The sargam differs from a key in that
it can be moved in pitch, until
the S is fixed.
-
-23-
There are as many striking similarities between the two genres
of music as there are
differences. Both systems have similar collections of pitches in
which a piece will (usually)
remain. Both highly value rhythm and melody, especially valuing
a singular instrument in the
foreground of the ensemble (though that instrument may change
throughout a piece). Both,
obviously, value both the original composition, variations upon
it, and improvisation within a
given structure. Their respective approaches to improvisation
reflect these comparisons.
1.3c: Comparison of jazz and Hindustani improvisation Before
engaging in a comparison of their improvisation styles, I want to
acknowledge the
differing meanings of improvisation globally. Ethnomusicologist
Nettl argues against the
Western dichotomy of improvisation and composition. In Western
music the distinction between
the two forms of creation is usually made based on the presence
or absence of written notation of
the piece. Nettl, in his comparative approach, explores the
multitude of cultures that do not have
a written notation system and yet produce music that the artists
themselves would not label as
improvised (Nettl 1974). For some, such as Persian music, the
performance of the dastgah is not
measured in notes and durations, but in the essence of what the
performer was trying to express.
Though a Western listener may hear each performance as two
different improvised pieces, a
native listener will experience the same essence, and thus the
same “composition.” Likewise, a
Hindustani performance of a raga has improvisational elements
throughout, but it is the essence
of the raga and the way it is expressed that usually remains
constant.
In jazz, however, composition and improvisation are much more
dichotomized. The form
of a typical performance is based off a composition, usually a
standard, that the big band or
combo will play as written. As stated previously, there is the
head (the melody) and the changes
underneath it (the harmony). The group will go through the head,
and then members of the group
-
-24-
will usually take turns improvising over the changes on repeat.
The number of solos played is
entirely up to the band, and can take anywhere from a minute to
an hour. To end the piece, the
members of the group will signal the front musician to play the
head again (usually this time
with more embellishments) and then the piece will end, perhaps
with a coda. Though the form is
quite dichotomized, there is some interplay between the
improvisational and compositional
moments of a performance. Each musician who plays a piece will
interpret the composition in
their own way, adding variations and embellishments to the head
(and/or the changes) as they see
fit. Likewise, many musicians draw melodic and rhythmic phrases
from the head during their
improvisations, using it as a home-base of sorts. Before this
experiment was conducted, each
participant was interviewed about their experience and
relationship to their music and own
personal craft. One of the jazz participants stated in her
interview that “the previously composed
melody is like a ground-base to lean on in improvisation.”
Though every jazz musician has their own method of
improvisation, many develop and
practice an arsenal of phrases, scales, and licks that they can
insert and incorporate into their
solo. Motivic improvisation is a technique commonly used by jazz
cats in which the musician
will introduce a motive (a short musical idea) and then develop
it, changing its length,
modulating up or down, inverting it, etc. etc. Though a solo is
technically improvised, not all of
it is created exactly in that moment1. When asked about the
thought process during
improvisation, one jazz participant stated:
1 For more in-depth insight into the thought process of a jazz
improviser, read Thinking in Jazz: the Infinite Art of
Improvisation by Paul Berliner (1994).
-
-25-
my ears are really reliable, so I most times go with them, but
there are patterns you
learn...you can dig through the patterns you already know and
use them...but I prefer to
go by what I’m hearing. After that, once I’m comfortable with
the chords and everything,
I try to build a little story around the improvisation...I try
to scat as if you’re having a
conversation.
While jazz improvisers must follow the harmonic changes,
Hindustani improvisers must
operate within the chosen raga. As stated before, the raga is
the melodic form that is trying to
express a feeling or color. When asked about the thought process
during improvisation, one
Hindustani participant said:
It’s all about the raag, you must focus on making the proper
note...you give preference to
the note, then comes the raag, then comes the composition and
everything...you immerse
in the raag, and everything else will come on its own.
A typical performance starts with the alap section, in which the
performer, usually a
vocalist, will introduce and develop the raga slowly and
deliberately. Then comes the jor, the
part in which the rhythm is established without the drum, and
the lead musician will perform
many variations of the raga. The jor gives way to the gat, the
fixed composition of the raga. The
tabla and subsequently the tala is introduced. After that, the
musician can improvise freely
within the raga and tala, analogous but not the same as the
changes and the time signature in
jazz. The musician can then return to the gat, speeding up and
intensifying until it ends in the
jhala, a fiery and dazzling section (Shankar 2009). Though the
Hindustani form is somewhat
reminiscent of the jazz form of theme → improvisation → theme,
the lines between the
-
-26-
composition and improvisation are much more blurred. The lead
musician focuses on the
tensions of the raga and the overall energy of the piece as it
unfolds, rather than following the
harmonic changes of a standard. For the purposes of this study,
I asked the Hindustani
participants to perform in such a manner that contrasted the
improvisation and composition
elements of a performance, so as to be comparable to the jazz
performance.
Though the parameters of improvisation differ between the
cultures, divergent thinking is
prevalent in both. When improvising, every note is a springboard
from which a multitude of
ideas can leap. In jazz, the directions it can go are mainly
dictated by the harmonic changes,
though modal improvisation can be and is widely used. In
Hindustani, the options lie in the
swara of the raga, and the relationship of each note to the
other. Regardless of the context, a key
element of improvisation is the rapid evaluation of multiple
options and the rapid and continuous
decisions between them; divergent thinking at hyperspeed.
In both groups, I asked the participants about the concept of
flow.2 Some participants
knew the concept, others understood once I explained it to them.
Overall, the jazz participants
had mixed experiences about flow, while the Hindustani
participants quite generally experienced
flow consistently. One jazz participant said this about
flow:
I’ve had it multiple times, like where you go offstage and I
really felt like something was
happening because of the reaction of the crowd, but I completely
black out...it’s as if part
of your brain goes numb, almost like the state you’re in while
you’re dreaming, you don’t
2 One item of import is for the majority of my participants,
English was not their first language (some were French, one was
Portuguese, and all of the Hindustani vocalists were Indian) and so
there may have been some miscommunications. However, I believe in
these excerpts the interaction was mutually understood.
-
-27-
have any control of it...but it’s still in the chords, you won’t
go out or anything, but it’s
not conscious…
[in order to experience flow] usually I have to be with people
and musicians that I’m
really comfortable with, people that value the role of the
singer...when you get musicians
who respect you and like what you do, they listen to
you...usually at that time I feel like I
have the space to create something.
When I asked the Hindustani participants the same question about
flow, one immediately
stated:
Yes, yes, always experiencing flow...it’s never preset, when
we’re about to start the
performance, previous to that my mind is completely blank, I
start fresh and go and build
on the raag… and every time it’s new….
[I experience flow] anytime where I sit to sing, at home or on
the stage, my mind will go
blank… it [the raga] will start on its own.
Overall, it did seem as if the Hindustani vocalists were much
more able to access flow
state, which as previously stated correlates highly with alpha.
This could be explained by
Pressing’s expertise model; as explained in more detail in the
Methods section, the Hindustani
vocalists on average have both been practicing their craft for
longer and practice more on a
weekly basis. The interactions with flow during improvisation
could largely be due to the
relative comfort with their art.
-
-28-
1.3d Comparison of the pieces: On the Sunny Side of the Street
and the Raga Yaman For the jazz section of the study, I chose a
standard that is relatively well-known by jazz
musicians: On the Sunny Side of the Street. Written in 1930 for
the Broadway show Lew Leslie’s
International Revue, it became a standard and has been covered
by the greats: Louis Armstrong,
Dizzy Gillespie, Billie Holiday, and Ella Fitzgerald, to name a
few. I chose the piece not only
because of the familiarity, but also because the harmonic
content is relatively simple and can be
theoretically reduced to its tonic key, which allows for
similarities in the way the vocalists could
approach the improvisation.
For the Hindustani section of the study, I chose the raga yaman,
a raga that is considered
to be one of the most fundamental ragas. Because the sangram is
quite easy and close to the
natural raga, most students of Hindustani music learn it at the
beginning of their studies. The
pitch content of the raga yaman is the same as the Western
Lydian mode (the same as the major
scale but with the raised fourth). The tala was the teentaal, a
very common 16-beat cycle.
As with all cross-cultural studies, I had to balance between
controlling for variables
between the subjects’ performance conditions and maintaining the
integrity of their process.
Though the usual performance of Hindustani music is much longer,
more elaborated and not as
dichotomized, this version still engaged the participants in
familiar material that was more
comparable to the jazz performance.
-
-29-
1.4: Objectives
I had two objectives for this study:
1) To determine the differing patterns of brain activity during
composed and
improvised music performances in both Western jazz musicians and
Eastern
Hindustani musicians.
2) To determine if there are different patterns of brain
activity between Western
jazz and Eastern Hindustani musicians when performing improvised
pieces.
1.5: Hypotheses
As mentioned previously, alpha oscillations have repeatedly come
up in EEG studies
focusing on both jazz improvisation and improvisational therapy.
Strong frontal alpha
synchronizations have been associated with flow state, and seem
to be a hallmark of
improvisation. Additionally, improvisation has been associated
with a disengagement of sensory
regions. I focused on alpha powers in the frontal and posterior
regions. As my experiment is in a
2x2 design, I compared improvisation to composed music within
each group, and then compared
improvisation and composed music respectively across groups.
However, as these scientific concepts of improvisation have been
derived within the
highly dichotomized improvisation-composed Western setting, they
may not be directly
applicable to the style in which Hindustani musicians perform. I
believe that the blur between
improvisation and composed music in the Hindustani tradition
caused the composed setting to
have improvisatory elements and its according neural
substrates.
-
-30-
Therefore, I hypothesized that there will be greater alpha
oscillation synchronization
during improvisation in comparison to the composed performance,
especially in the prefrontal
and posterior regions. I believed the distinction will be more
evident in the jazz group than the
Hindustani. It follows that between the composed elements, the
Hindustani group will have
proved to have more alpha oscillations involved than the jazz
group.
Because of the lack of harmonic changes in the Hindustani group,
I believed that the
Hindustani participants will have had even less sensory
activation during improvisation than the
jazz participants during improvisation. Between the composed
portions I hypothesized that there
will be higher levels of alpha in the Hindustani participants,
as they incorporate more
improvisatory elements in their compositional sections.
If alpha oscillations are prevalent in the prefrontal and
posterior regions, we can
reasonably infer that improvisation incurs a state of
relaxation. If there is more alpha prevalence
in Hindustani music, it not only implies that their
improvisation would be adequate for music
therapy, but perhaps even more appropriate than jazz
improvisation.
-
-31-
2: Methods I conducted my study in two parts, first with jazz
vocalists in Montreal, Canada, and
second with Hindustani vocalists in Bangalore, India. Each part
dealt with different musical
content, but the experiment design remained relatively the same.
Henceforth, the first part of the
study will be referred to as the jazz group, and the second part
of the study will be referred to as
the Hindustani group. Each participant went through the study
individually.
2.1: Participants
I had four jazz vocalists in the jazz group and four Hindustani
vocalists in the Hindustani
group, totaling in eight participants. All eight participants
were female. Only one jazz participant
had a history of mental/psychiatric disorders (depression), and
none of them had a history of
brain injury nor drug/alcohol abuse. All were in good health and
could read and speak in English
well. The jazz vocalists were recruited through a vocal teacher
at the University of Montreal and
the Hindustani vocalists were recruited through Dr. Shantala
Hegde, clinical psychologist and
researcher at NIMHANS (as well as a Hindustani vocalist
herself). All participants were
informed about the study prior to consenting, in line with the
requirements of the Institutional
Review Board at Brown. They were each compensated with a
handwritten note and a Brown
University pen.
Prior to the study, each participant was asked to sign an
informed consent form and fill
out a demographic questionnaire about their practice and
performance practices. All forms are
available in Appendix B. Being that the population size is small
and their demographics are
varied, the averages and deviations from the mean differ widely
between the two groups. A
breakdown of participant demographics can be seen in Table 2.
There are some items of interest
between the two populations that are in line with cultural
practices around the two genres of
https://drive.google.com/a/brown.edu/folderview?id=0B9q7T6jIrQ_udGkyM29sbnBGZTg&usp=sharinghttps://docs.google.com/a/brown.edu/document/d/1MJic9tzZCo86sXHKfqaZktXxZSJfLgfEWsxSEHSYChA/edit?usp=sharing
-
-32-
music. The age at which most jazz vocalists start studying the
form is in the late teens/early
twenties, while it is common, if not expected for Hindustani
vocalists to start around age 4. This
explains why the average age between groups is similar and the
average years of experience in
their field is so drastic, with the Hindustani vocalists
averaging about 10 more years of
experience than the jazz vocalists. Overall, the Hindustani
population practices more frequently
and for longer than the jazz population does, but the jazz
population tends to perform more
frequently per month than the Hindustani population. For more
information, refer back to section
1.3c for narratives about the participants’ interaction with
their musical genre.
Mean Range
Age Jazz 32.25 23-51
Hindustani 35.50 29-48
Years studying voice Jazz 16.25 6-31
Hindustani 26.75 13-40
Years studying improv Jazz 12.25 6-25
Hindustani 15.38 1.5-25
Practice sessions per week Jazz 5.38 4.5-7
Hindustani 6.50 5-7
Hours per practice session Jazz 1.75 1-2.5
Hindustani 2.19 0.5-4
Performances per month Jazz 5.88 2.5-12
Hindustani 1.08 0.85
Table 2: Participant Demographics. This table shows the mean and
range for demographics between jazz and Hindustani participants.
Though comparable in age, the Hindustani participants have on
average been studying for longer and more per week. However, the
jazz participants perform more frequently than their Hindustani
counterparts.
-
-33-
2.2: Materials
I recorded audio, video, and EEG of each participant’s
performance. I used the Zoom
H4N Portable Recorder for the audio, the Canon Vixia HF G10 for
the video, and the Emotiv
Epoc EEG headset for the EEG. I used the iReal Pro app (Technimo
LLC) to provide piano,
bass, and minimal drums for the jazz vocalists and I used the
iTabla Pro app (Prasad Upsani) to
provide tanpura and tabla for the Hindustani vocalists. All of
the data was backed up in a secure
folder on my personal laptop.
The Emotiv system is a research-quality system3 with fourteen
channels (AF3, F7, F3,
FC5, T7, P7, O1, O2, P8, T8, FC6, F4, F8, AF4 within the
international 10-20 system) and two
CMS/DRL references at P3 and P4. It has a sampling rate of 128
SPS via single ADC sequential
sampling. It uses a low-pass sinc filter with notches at 50Hz
and 60Hz, leaving a frequency
bandwidth of 0.2 - 43Hz. It is a dry contact system, so the
electrodes had only to be wetted with
saline solution prior to setting it on the participants’ heads.
The headset uses Bluetooth to
connect with the Emotiv software, TestBench™. TestBench™
software provides real-time
display of EEG, contact quality, FFT, gyro, and marker events. I
imported the EEG data from
TestBench™ into EEGLab, an EEG analysis toolbox for MATLAB
(R2013b; The MathWorks,
Inc.). EEG is the recording of electrical activity along the
scalp; it is useful for examining
changes in power of frequencies over time and at particular
moments in time. The data was
preprocessed in a number of ways. I ran the data through a high
pass and low pass filter at 5 and
40 Hz respectively. I removed the baseline to average the raw
data, and ran independent
component analysis to correct for artifacts, and eliminated
channels with excessive artifacts or
poor contact quality.
3 The company states that it is research grade; however, this is
contested by some studies and is addressed in the Discussion
section.
https://emotiv.com/product-specs/Emotiv%20EPOC%20Specifications%202014.pdf
-
-34-
2.3: Protocol
Both groups underwent identical procedures, save for the musical
content.
Prior to the study, I emailed each participant with information
about the study and with the
musical piece to prepare. For the jazz section, I sent them the
chart and recording of the changes
to which they would sing in the actual study. I asked all of the
jazz participants to perform the
tune “On the Sunny Side of the Street” in G major at a tempo of
140 beats per measure. As I was
unsure as to how familiar each participant was with the
standard, they were given the sheet
music and backing instrumental ahead of time so they could
prepare. The backing instrumentals
were synthesized piano, bass, and drums from the iReal Pro app.
The piece was in G major at a
tempo of 140 beats per minute. The recording went through the
form (the changes) three times in
total. The chart and the recording can be found in Appendix B.
For the Hindustani section, I
asked them to select a chota khyaal in the raga yaman. As this
raga was well-known by all
participants, I gave them no preparatory material. For the
background music, I used the tanpura
and tabla sounds in iTabla Pro, also in an analogue to the key
of G at a tempo of 140bpm. The
participants performed the first part with the drone alone, and
then added the tabla themselves
when ready by clicking on the app. I asked participants in both
groups to engage with the
material for at least an hour prior to the study, so that all
participants could enter the study at a
similar level of familiarity.
Upon entering the study, I gave them a verbal overview of the
study and gave them the
informed consent form, which they read and signed. They also
filled out the demographic form
as mentioned above. As I set up the recording devices, I gave
them some time to warm up and
practice, if they so chose. I then conducted an informal
interview with them, asking about their
experience with and relationship to performing, improvisation,
and music in general. I then
arranged the Emotiv headset on their head and adjusted the
position and/or saline solution of
https://drive.google.com/a/brown.edu/folderview?id=0B9q7T6jIrQ_udGkyM29sbnBGZTg&usp=sharing
-
-35-
each electrode to optimize impedance levels. Then I started the
accompanying music and asked
them to perform while I recorded audio, video, and EEG.
For the jazz section of the study, I asked them to sing the head
once all the way through,
then improvise over the form once all the way through, rest for
two As, then come back in with
improvisation for the last BA. For the Hindustani section of the
study, I asked the participants to
perform the aroha and avaroha (the ascending and descending
scales) of the raga yaman in aa
kar (without the swara syllables, instead with an “aah” sound).
After that, I asked them to sing a
chota khyaal (a short composed portion) without too much
variation. Then I asked them to
engage in batath (improvisation) for about 3 minutes or so.
Then, as with the jazz section, I
asked them to rest for about 30 seconds, then proceed with more
improvisation until they saw fit.
I allowed them to choose when they moved through the form, as it
was more in line with their
practice. The performance in the jazz section was about 3
minutes, while the performance in the
Hindustani section was about 7-9 minutes. In total, including
set up and break down, each study
took about 60 minutes.
-
-36-
3: Results
3.1: Preparation and selection of the data
I analyzed frontal and posterior regions. For the frontal
region, I averaged channels AF3,
F4, and AF4. For the posterior region, I averaged P7, O2, and
P8. Channel F3 was defunct prior
to the study, and Channel O1 broke between the conduction of the
jazz section and the
Hindustani section. Channel AF4 had poor contact quality for one
jazz participant and one
Hindustani participant, so those were removed for the analysis
of just those participants. One
jazz participant also did not have usable data from P7, so that
too was removed from the
analysis. The system lost contact entirely halfway through one
Hindustani participant’s trial, and
so that entire trial was excluded from the analysis. Figure 1a
and 1b shows the channel locations
and a visual representation of each participant’s channel
contact quality.
Figure 1a: Channel map of the Emotiv system in accordance to the
International 10-20 system. AF3, F4, AF4, P7, O2, and P8 were the
channels used in this study.
-
-37-
MJ001 MJ002 MJ003 MJ004
BH002 BH003 BH004
Figure 1b: Channel contact quality for each participant. The
jazz participants are in the top row, Hindustani participants in
the bottom. BH001 is removed because their trial was rendered
defunct by the headset.
I examined the alpha oscillatory patterns in the participants,
looking at specifically at the
power of the alpha band (7.5-12.5 Hz) in the frontal and
posterior regions. I made 16 second
epochs of three conditions of the study: composed, improvised,
and listen. In order to make the
participants’ data comparable to one another, a baseline from
which the conditions are contrasted
needs to be established. For instance, one participant could
have abnormally high alpha in
general, and so their data could pull the rest of the group up
unnecessarily. An ideal baseline
would be EEG data gathered while the participants were at rest.
Unfortunately, I did not collect
such data, and so I used the listen condition as the baseline
for the alpha band instead. From that,
I constructed a spectrogram for each condition (jazz-composed,
jazz-improvised, Hindustani-
composed, Hindustani-improvised) in each region and calculated
the average alpha over time,
also with the averaged baseline subtracted.
-
-38-
3.2: Descriptive Analysis
Unfortunately, the sample size of this study was too small to
draw any statistical analysis
from the data, so instead the analysis will be purely
descriptive. Spectrograms (or time-frequency
plots) of the respective frontal and posterior regions of the
subjects in each condition are shown
in Figures 2 and 3 below. This shows the power of frequencies
5-40Hz as they change over time
in each 16s epoch. I am specifically focusing on the alpha band
to see if there are any significant
changes that may correspond with events in the music.
3.2a: Spectrograms
Figure 2: Spectrogram of frontal region. Each spectrogram shows
the frequency powers for the first 16sec of each condition in the
frontal region (channels AF3, AF4, and F4). This data is plotted
relative to the baseline made from the listen condition. Black bars
mark the boundaries of alpha (7.5-12.5Hz). A is the composed
condition in the jazz group, B is the improvised condition in the
jazz group, C is the composed condition in the Hindustani group,
and B is the improvised condition in the Hindustani group.
-
-39-
Figure 3: Spectrogram of posterior region. Each spectrogram
shows the frequency powers for the first 16sec of each condition in
the posterior region (channels P7, O2, and P8). This data is
plotted relative to the baseline made from the listen condition.
Black bars mark the boundaries of alpha (7.5-12.5Hz). A is the
composed condition in the jazz group, B is the improvised condition
in the jazz group, C is the composed condition in the Hindustani
group, and B is the improvised condition in the Hindustani
group.
In the frontal region, the composed portion of the trial showed
periodic strengths and
weaknesses in alpha levels for the jazz participants. There were
no apparent trends, save for an
initial burst of alpha strength at the 2 sec mark. The
Hindustani participants also shared an alpha
prevalence in the composed section, followed by a marked
depression in alpha power at the 3.5-
5.5 sec mark. After that, they had relatively consistent alpha
strength. For the improvised section,
the jazz participants had relatively even alpha powers
throughout, with slightly stronger values
-
-40-
initially. The Hindustani participants had overall stronger
alpha values in the improvised section,
with a depressing at the 8-10 sec mark.
The posterior region showed only slight differences from the
frontal region. In the jazz
composed section, there was a stronger initial burst of alpha at
2 sec, with less evenly distributed
alpha throughout. The Hindustani composed section had a less
evident but still present
depression at the 3.5-5.5 sec, with a novel burst at 11sec. The
jazz improvised section also has
less evenly distributed alpha, with more evident bursts at
10.5-11.5 sec and 14 sec. The
Hindustani improvised section had much more contrast than its
frontal counterpart, with more
alpha everywhere save the 8-11 sec mark.
3.2b: Averaged Alpha To more easily compare alpha powers between
groups and conditions, I averaged the
alpha power over time. Below are Figures 4 and 5, graphs of
alpha averaged compared to the
baseline over time for the frontal and posterior regions
respectively.
Figure 4: Frontal Alpha Averages Figure 5: Posterior Alpha
Averages
Overall, all of the alpha averages were negative relative to the
baseline, which implies
that there was an overall reduction alpha amplitude relative to
the baseline. However, in the
frontal region both the jazz and Hindustani groups had higher
alpha power in the improvised
section than in the composed. The Hindustani group had much less
alpha in the composed
-
-41-
section than the jazz, but had higher levels of alpha in the
improvised section in relation to jazz.
Posteriorly, the Hindustani group still had higher alpha power
in the improvised section than the
jazz group, but both were significantly lower than in the
frontal region. The alpha powers for the
composed section were essentially the same for both jazz and
Hindustani groups. Interestingly,
though the values were different, the jazz and Hindustani
participants shared the same
relationship across brain areas; the jazz participants had
higher alpha power in the composed
condition while the Hindustani participants had higher alpha
power in the improvised condition.
-
-42-
4: Discussion
4.1: Discussion of the results
This study aims to use alpha power in the frontal and posterior
regions as a lens onto
divergent creative thinking (specifically through improvisation)
between two cultural groups,
jazz and Hindustani musicians. The data was partitioned into
sixteen second epochs for three
conditions: composed, improvised, and listen. The alpha power in
listen condition were used as a
baseline to which the composed and improvised sections were
compared. Ideally, a proper
baseline would have been one in which the participants were
completely at rest; instead, this
baseline was drawn from a condition in which the participants
were not actively performing, but
were processing the background music and perhaps thinking about
their next musical decision in
the subsequent condition. This means that the baseline is not
static, and so this data must be
taken provisionally. Because the results were drawn in relation
to this baseline, it is difficult to
determine whether the outcomes are due to the actual conditions
themselves or due to the
dynamism of the baseline.
As stated previously, it was difficult to compare conditions
exactly to one another across
cultural groups and practices. One large difference between
groups in the experimental design
was that while the jazz participants all sang the same composed
piece, the Hindustani
participants performed a chota khyaal of their choosing within
the raga yaman. That means that
although inferences can be drawn from the changes in alpha power
over time for both groups,
only the data from the jazz group can be specifically
synchronized to a certain point in the music.
Letting the Hindustani participants choose their composed piece
also adds a variable of
complexity; it is harder to determine what parts were harder for
the group to engage with overall
when the compositions may have different periods of
complexity.
-
-43-
It is a point of intrigue, too, that the Hindustani group showed
more apparent periods of
lower alpha than the jazz group did. Both groups did have an
initial onset of higher alpha levels
in the improvised condition. This could imply that the initial
moments of improvisation come
naturally, and once the first motive is played out the next
choice must be made, which requires
more executive function.
The finding that all of the averaged alpha powers were negative
relative to the baseline
was not in line with the hypothesis of significant alpha
activation during improvisation. This may
be due to the nature of the baseline; the participants may have
had high levels of alpha while
listening to the music and so the data drawn in relation to that
would be negative. The low alpha
powers may also be due to the experimental conditions; the
participants were performing in
circumstances quite out of the ordinary. Many of the
participants appeared to be nervous
(especially under the uncomfortable EEG headset), and so they
may not have been relaxed
enough to enter flow state.
However, in comparing the groups and conditions to one another,
we come across
observations that support the hypotheses and observations that
contradict it. In the frontal region,
both jazz and Hindustani participants had higher alpha powers in
the improvisation condition
than in the jazz. This is in line with the hypothesis that
improvisation, a divergent thinking task,
will elicit higher levels of alpha than performing previously
composed music. However, the
distinction between alpha power in the composed and improvised
sections was greater in the
Hindustani group than in jazz, unlike what I had previously
hypothesized. Additionally, the
Hindustani participants had much lower alpha power overall in
the composed section than the
jazz participants did. Both of these observations undermine the
hypothesis that the lack of
separation between improvisation and composed music in
Hindustani tradition would lead to
-
-44-
higher and more comparable powers of alpha in the Hindustani
group. However, this may be
explained by the sheer complexity of Hindustani vocal
performance over jazz. Though it is true
that the distinction between improvisation and composed music is
much more blurred in
Hindustani, the way in which the musician must adhere to the
structure of the raga and develop
tonal relationships is much more demanding than the way (most)
jazz vocalists develop the head
of a tune. This is evident in the practice of Hindustani music.
As mentioned previously, most
Hindustani vocalists start learning their craft at age four, and
will practice for multiple hours
every day of the week. They will learn a couple of ragas and
work with them for years on end.
Jazz vocalists, on the other hand, tend to start much later and
practice much less. They also tend
to value a breadth of musical knowledge over depth by learning
as many standards as possible,
so as to be flexible and dynamic in their performances and jams.
A transcription of an excerpt
from one of the Hindustani participants can be examined in
relation to the chart for the jazz tune
to see the differences in melodic complexity. This can be found
in Appendix B.
The alpha power in the composed condition for both groups
remained relatively the same
across frontal and posterior regions. Both groups had much lower
alpha power during
improvisation in the posterior region. This is a trend that has
been found in multiple studies, in
which there were patterns of decrease from anterior to posterior
regions while doing divergent
thinking tasks (Jauk et al. 2012, Fink et al. 2009). Although
both groups saw lower alpha powers
posteriorly during improvisation, the jazz and Hindustani
participants experienced the same
relationship to each other across conditions and brain
areas.
Though the overall negative power of alpha across the board is
not in line with my
hypothesis, the comparisons between jazz and Hindustani
participants during composed and
improvised conditions do in fact support the theory of frontal
alpha-prevalence during divergent
-
-45-
creative processes such as improvisation. Not only did both
groups see higher alpha powers in
the improvised section than the composed, but the Hindustani
group showed higher alpha power
in improvised than the jazz group did. This corresponds to the
participants’ narratives; when
asked about flow, the jazz participants cited particular
circumstances under which they
experience flow, while most of the Hindustani participants
stated that they felt flow almost every
time they performed. As stated before, the average Hindustani
vocalist starts learning their craft
much sooner and practice more regularly than the average jazz
vocalist. This easier access to
flow can be connected to Pressing’s (1988) expertise-related
model of improvisation; the more
experience one has, the more automated the lower-level processes
are, and the more likely one is
able to experience flow.
Earlier on I addressed the role of musical improvisation and
health, specifically its role in
music cognition. I posited that higher alpha powers in the
Hindustani improvised condition over
jazz may imply that Hindustani improvisation may be more
suitable for music therapy, as alpha
indicates relaxation. Though the results did show a higher alpha
power in the Hindustani group
than in jazz in the improvised condition, it appears that the
source of the alpha difference is from
years of training and experience that the Hindustani
participants have, and so though they may be
more relaxed in the improvised condition, the amount of work
needed to achieve that state is
unsuitable for a therapeutic condition. However, this does not
preclude non-Western forms of
music being used as tools in music therapy. Usually the
improvisation in Western therapy is free-
form and not evaluated based on aesthetic; this means that any
form of music, Western or no, can
act as a foundation in which patients can play without
restrictions.
-
-46-
4.2: Limitations of this study
It would be remiss to not examine this study critically. Though
this study was designed
with the guidance of many qualified people, there were
impediments to the design and execution
that should be addressed and changed in subsequent studies.
The largest issue this study faced was the small sample size.
Four participants per group
with data from one of the Hindustani participants defunct left
seven total trials, a dataset
insufficient for statistical analysis. Not only was the data
small in terms of sample size, but in the
number of available channels as well. The Emotiv EPOC EEG
headset, while very attractive in
cost and portability, had a limited number of channels and a
fragile construction. The headset’s
connection to the computer was unreliable, and two of the
electrodes broke during the data
collection. The Emotiv headset is marketed as research grade,
but as Duvinage et al. (2013)
found in their study testing the headset’s ability with an P300
ERP trial, the headset is more
suitable for non-quantitative activities, such as
videogames.
Another source of imprecision was the subjective event marking.
The markers indicating
the onset of each condition were placed post-trial by
synchronizing the video and EEG data by
eye. The markers would have been much more precise if they were
placed by the primary
investigator in real-time during the trials. Another flaw in the
experimental design was the
location of the trials. In total, there were four different
locations in which the study was
conducted: one in Montreal and three in Bangalore. The changes
in location in Bangalore were to
accommodate the participants’ schedules, and varied from a
bedroom to a practice room to an
EEG lab. Each location had differing levels of distractors, and
only one location had a Faraday
cage to reduce electrical noise; this led to differing artifacts
per trial. To conduct every trial in
one location would have removed many artifacts from the
data.
-
-47-
Aside from location, there were many inconsistencies between
groups. The conditions
themselves varied: in the jazz group they were all asked to sing
one particular song, while the
Hindustani group was asked to choose a composed piece within one
particular raga. This added
an unnecessary variable to the Hindustani group. Additionally,
the Hindustani group did not have
a specific time frame for each condition, while the conditions
for the jazz group were temporally
dictated by the form of the song; this makes temporal
comparisons much more difficult to make.
Though the jazz participants knew more naturally when to switch
conditions, they were not
nearly as familiar overall with On the Sunny Side of the Street
as the Hindustani participants
were with their chota khyaals within the raga yaman. The raga
yaman is one of the simplest and
earliest taught ragas in the Hindustani tradition, and so the
participants were very well versed in
it. While some of the jazz participants did know the tune,
others were reading the chart as they
sang, which caused many eye movement artifacts in the data. Most
of the Hindustani participants
had their eyes closed for the majority of the trial, which has
been shown to produce a marked
increase in alpha (Barry et al., 2007). Another difference
between the groups was the syllables
used during improvisation; the jazz participants all used their
preferred scat syllables, and the
Hindustani participants switched from aa kar to syllables of the
swara to even words from the
chota khyaal. Language and music share an intimate relationship,
and the utilization of real
words over non-real words with implicit value (scat syllables)
over general sounds can influence
which parts of the brain are active during improvisation.
Of course, an issue that must be considered for any type of
music cognition study is how
to strike the balance between controlling variables and
inhibiting a creative environment. Music
in its natural state is often messy, with many moving parts and
events. While studying the neural
substrates of lower-level psychoacoustic or musical processes
may be done in isolation, it seems
-
-48-
that such a cognitively complex function such as improvisation
is hard to take out of context.
Limb and Braun (2008), among others, support Burgess et al.’s
argument for an “ecologically
valid” model in which the study environment is conducive to
accurate reproductions of these
complex functions (2006). Nonetheless if too many variables are
involved in the data, it makes it
much more difficult to draw conclusions to specific causes. It
is then the responsibility of the
investigator to reconcile the two needs of the study in a way
that is both conducive to the creative
process and cogent in its conclusions.
-
-49-
4.3: Suggestions for future studies
In the context of the limitations to this study described above,
below are
recommendations for future studies.
Use a larger sample size. This will allow the general body of
data to withstand fallout
and allow for robust statistical analysis.
Use a research grade EEG system with at least 32 channels.
Utilizing a more reliable
system with more channels will not only prevent crucial data
from being lost, but will also allow
for spatial analysis and functional connectivity analysis. A
higher grade system will also be
conducive to adding event markers during the trial.
Conduct all trials in the same space. Though it would mean
removing one of their
groups from their home culture, I would recommend conducting
each trial in the same location,
preferably a room equipped with a Faraday cage to reduce
external electrical noise. That would
help reduce artifacts and eliminate external
distractions/variables.
Add a rest section to the conditions. Having some EEG data while
the participants are
at rest will provide a much more reliable and stable baseline
than my use of the listening
condition provided.
Restrict the Hindustani group to a single chota khyaal. This
will allow the investigator
to synchronize commonalities across participants. I would
recommend finding a piece that is
comparable in length to On the Sunny Side of the Street (or
another jazz standard with simple
harmonic changes). It would also be beneficial to ask the
Hindustani participants to improvise
for the same time as they performed the chota khyaal, so as to
regulate the timing of each
condition.
Match the participants in experience and familiarity with the
piece across groups.
Though it may be difficult, try to find participants with
similar years of experience practicing
-
-50-
their art. That will allow both groups to be compared at the
same caliber. Additionally, try to find
participants who are all equally familiar to the piece they are
performing. It could be helpful to
give each participant an opportunity to learn (either on their
own or with guidance) the piece two
weeks prior, and require a certain amount of practice hours
prior to the trial. That way the
participants could engage with the material equally, and not
have to divert their attention to a
sheet of music.
Explore additional hypotheses and methods of analysis. As stated
above, using a
higher-grade EEG system with more channels would allow for more
spatial analysis. It would be
of interest to see exactly where alpha is prevalent, and how
that may change over time.
Additionally, looking at the functional connectivity between
frontal and sensory regions would
help address the relationship of lower-level sensory input
processing to upper-level executive
decision making and how that relates to flow (see 1.1 for more
details). Because so much of the
improvisation literature is measured by fMRI, a future study
would benefit from using both EEG
and fMRI methods and comparing the results to each