Anda di halaman 1dari 36

Brain, Mind, and Cognition

This is your brain on music by Daniel J. Levitin


Essay
Joel Edenberg

First of all I think this book could be a lot better if it was presented in a more interactive
format. Majority of the time when I was reading I did not have access to any media player.
So I could not actually listen to the songs mentioned in the text. As a lot of the mentioned
songs where unknown to me, or at least based on the titles, the general understanding of
contexts was probably suffering. An interactive computer version with links to music
samples could offer a lot better overall experience.
But in general I enjoyed the book and found it interesting. It contained a lot of interesting
facts and some pretty unique ideas. Even though I actually have studied music before I found
the explanations in first chapters really helpful. Sometimes the heavy usage of people names
got a bit annoying but other than that it was good.

The most interesting though or idea
I found the explanation why we enjoy the music the most interesting. It made me think
about something I had not thought about before. Why do we like music and what kind of
music we like? Before reading this book I considered this question too abstract to actually
answer. Similarly we could ask that why do you like this specific color or why do you like that
car? The answer seems to lie deep in persons personality and coming up with a reasonable
explanation seems hard.
The author states that we like music because it plays with our expectations. When we hear
music our brain tries to predict what sound we will hear next. Usually these predictions are
met as the thing we call music has to have some kind of structure or sound pattern. But
sometimes those predictions are violated and we hear something unexpected. These
surprises keep our mind interested in the given music.
Now why would we find listening to predictable patterns pleasurable? If we consider the
general concept of how brains work, presented by Jeff Hawkins, we know that predicting is
the core working principle of brain. So our brain is constantly predicting everything that is
happening to us. While performing activities we are constantly improving our predictions as
we get additional information. The initial guesses are often very rough approximations and
end up being changed a lot refined over time. In general predicting correctly is pleasurable
because our brain has successfully figured out the underlying patterns. We feel good as our
brain has succeeded. But after some time it gets boring and fictional - does not resemble real
life. It seems like an abstract world with ideal concepts. There are no unexpected events to
make the situation more challenging.
Now if we are listening to some very simplistic music all of our predictions are met. In real
life that is never the case. So in order to keep the music interesting there have to be some
violation of expectations. The music has to take unexpected turns to keep us interested. Our
brain is forced to focus more on the music, because it keeps getting the next sounds
wrong. It tries to come up with new schemas and patters for describing the sounds it is
hearing but every now and then the model is violated by the music composer.
But why would we enjoy predicting wrongly? Why should we feel any kind of positive
feelings when our brain fails at its core functionality predicting. I think it has something to
do with the cost of actions. In real world every action we take has a cost. Imagine a person
walking - for every step our brain predicts where to put the feet in order to complete the
next step. Based on the visual information from eyes brain predicts where to place the feet.
Now if this prediction is off by even a small margin (ground is actually lower than we thought
it will be) we are forced to waist a lot more resources - energy. We try to avoid falling down
by using a lot of additional muscles. We use arms to regain the balance, strengthen the
entire body and try to re-adjust the foot position rapidly. So every time our brain
expectations are not met we get penalized with energy loss.
Now listening to music is in some way different. There is no penalty in predicting the next
sounds wrongly. It is like a harmless game where we do not pay any fee for performing
badly. I think this is one of the reasons why we find small anomalies in music pleasurable.
While listening to music our brain does not receive any negative feedback for being wrong.
Based on these ideas we can start to explain as well why people like specific types of music.
The patterns presented to the listener need to be simple enough for recognition but
unpredictable enough to keep the listener interested.

Ideas obtained from the book for building intelligent systems:
Music is closely related with emotions. It can change our mood interact with us. But giving
emotions to machines is probably useless and could even lead to some unwanted results. So
I do not think giving artificial intelligence the ability to understand music is necessary.
Without emotions there is no beneficial value between noise and what we call music. Sure
machine could find patterns in sounds as well but without emotions there is no rewarding
value or purpose for it.
What we could do instead is to learn from our auditory system how to handle sound. The
input of sound is usually very noisy containing a lot of sound sources mixed together. The
problem of identifying sound sources and separating them can be solved by using similar
techniques to how humans do it. For example the distance could be determined by the
volume of sound and the relative position could be determined by comparing 2 sound inputs
from different locations. While even humans have trouble with some specific sound
identification tasks (for example determining whether the sound is coming from front or
from behind) we can use similar solutions to what evolution has come up with. Thanks to
outer ear some higher frequency sounds are cut off if they are received from behind. But if
we cannot predict if those higher frequencies should be present then just slightly turning our
head can identify the position. Closely learning how sound is processed by humans can lead
to new solutions for artificial intelligence.
Also the way how music and songs are stored in our memory could be beneficial to study
and understand. We remember the abstract representation or schema of different attributes
describing each song, but at the same time we can remember specific facts about it (like
lyrics, in what key it starts, tempo, etc.). By recalling one aspect of the song we can recall all
the other variables. This highly intertwined data structure would allow very flexible queries
and data retrieval.

1

This is Your Brain on Music: The Science of a Human Obsession
Review by Artem Kaliuk

If you develop an ear for sounds
that are musical it is like
developing an ego. You begin to
refuse sounds that are not musical
and that way cut yourself off from
a good deal of experience.
John Cage (1912 1992)
I do not know many people going for higher education in their thirties. Among those I
know (or those I have only heard of) I cannot recall anybody making not just a
successful career, but a universally-recognized name in science. This is why Daniel
Levitin already seemed to be an interesting person even before I started reading his
book. Having provided a detailed overview of a music theory in the first chapters of the
book, Professor Levitin steps into discussion of the brain predicting, reacting and
delighting us with music.
The theory behind the text lines resembles with the layered prediction-memory
framework which J eff Hawkins introduced in his book (here the top-down and bottom-up
cases are explained by top-level and low-level processing, respectively). Accepting the
importance of physiological structure of the brain to the cognitive processes, Levitin,
however, emphasizes several times on his particular interest in mind as an abstraction
of the brain activity. Giving an analogy with software burnt into the hardwired structure,
the author tries to see the reasons for feeling and enjoying the music. As before, I
disagree with such analogy as software instructs and controls the hardwire, while mind
is something that comes out as result of electrochemical (say, physical) activity in the
brain. Our minds can be adapted by simply affecting this activity (with drugs or other
stimuli), while the software stays either running properly or erroneously due to internal
failures or memory damages. Mind is based on the current physical activity and
previous experience. I can only accept the notion of software applied to genes: the
genetic information which we receive from our parents to some extent defines our
temper, preferences and future development. But it is still a rough comparison to a
computer program what we receive encoded in genes does not necessarily define our
fate (children having aptitude for music being exceeded by their less-talented but more
hardworking fellows is an encouraging example provided by Daniel Levitin).

2

Fascinating in this book is:
- that studying music or how the brain works on music allows us to look incredibly
farther into the brain than it seems at the first glance. Music proves to be a good test
tool involving prediction, emotions and grouping.
- the newly-discovered complementary role of cerebellum. For a long time the oldest
part of the brain was thought of as a control room for human motions. Now there is
evidence that cerebellum also stores the temporal information about the music we are
listening to and then it recalls this information when while reproducing the music.
Moreover, an amazing fact is that cerebellum was discovered to be a center for
emotions;
- categorization. In previous readings we were already exposed to an amazing
paradigm of invariant representations in our memory. In This is Your Brain on Music the
author explores the idea of invariant memory as a categorization problem. Here we
witness a debate between two classical groups of memory theorists constructivists,
who stand for idea of some general low-detail pattern being stored for each experience
we encounter, while the details are defined and added later; and record-keepers who
insist on storing exact details of each object or event just the way it is done by a tape-
recorder. At the first glance constructivism seems to be a way more reasonable
approach, but Levitin breaks our expectation by giving examples which prove the two
points of view to be both right and wrong. As a result, we are introduced to a new
memory theory. Professor Levitin assumes that with multiple-trace memory every
experience is stored and, what is more important, can be retrieved and replayed in our
minds by firing the right cue a complementary experience (a melody, an odor or a
friends joke) which we had at this moment. That is why we sometimes laugh when
somebody makes a statement which is not funny by its nature (but causes us to fire a
neural network storing an inside-joke related to this particular phrase); that is why we
want our professors to give us interesting and easy-to-understand examples when
explaining a highly-complex theory. The more we access the memory (or, simply
speaking, the more we repeat a particular primary experience in a tandem with the
complimentary one), the stronger the synapses become and the faster the information
retrieval will be done. This is already a good motivational example both for chip memory
designers (and I believe similar optimization strategies are already implemented) and
teachers (to make the students remember things in an easy way. However, education
still possesses a huge potential for improvement).
- chapter about emotions. During our previous meetings several times we touched
upon the issue of having an artificial mind without emotions. There is a hard-to-observe
mechanism behind emotions when we listen to music. However, it was shown that at
least three components of the brain are involved into emotional reactions to music
3

(amygdala, cerebellum and nucleus accumbens). Even more interesting is to see when
we experience pleasure: either when predicting the next piece of music correctly or,
when a musician surprises us by violating the structure which we were expecting.
Indeed, being a good musician requires not only having the technical skills, but also
breaking the patterns to call emotions (why do you think Nickelback is the worst band
ever?). Surprisingly, this phenomenon can be discovered in the fields not directly
related to music: breaking the clichs (for instance, in product design) is one of the key
success factors in business.
Inspiration for building intelligent systems
I am quite sceptical about the idea of creating a silicon chip with a voice of Christina
Aguilera (however, mounted on top of a dancing robot, it could be a perfect participant
for the Eurovision song contest), so I did not consider any androids for this part. Instead,
reading about the categorization for MP3 search engine made me recall some thoughts
I had several years ago. The author mentioned problems of finding the versions of the
same song by different artists. It could be useful to apply the speech recognition
techniques to extract the lyrics of the song. As each artist has a distinct timbre, it would
be interesting to analyze the timbre of the voice (in the way humans do) on each
uploaded song based on a small fragment and create a category for this particular artist.
Based on the recognized text we could now look up for the song Mr. Sandman; based
on the difference in voice timbres, it will give out versions by different artists.
Creating an artificial auditory system as a functional replica of the human one can be
widely used for sensor fusion. Most of the current applications in robotics involve
processing visual and vestibular data from cameras and accelerometers. Analyzing the
audio data coming from the environment can help significantly when detecting obstacles
or moving threats for both human and robot navigation applications (inside the factory
building, on the road or even in the air). It was mentioned that visual and auditory
systems are the best developed. It would be also interesting to apply similar processing
techniques (recall that cortex regions for each sensory system are all of the same
structure and functional principle) for haptic interactions and odor analysis.
After all, I cannot restrict myself from the negative emotions I had about the book:
Professor Levitin LOVES referring to numerous people he met, talked to, had a drink
with or was just ignored by. I consider this constant name-dropping together with a deep
excursion into the music theory (which I find much less useless and annoying) to be one
of the reasons for my slow progress through some of the chapters (I got simply lost in
names and DAH-dah-ta DUM-dums). Despite that (and the authors criticism of Frank
Sinatra), this book made me think of a music in a different way and it is quite likely that I
may read it once again (and I have already recommended Your Brain on Music to
some of my friends).
This is Your Brain on Music, by Daniel Letvin
Scott Kenealy
January 7, 2012
1 Introduction
Daniel Letvins book, This is Your Brain on Music, is best thought of as a series of essays related to the
role of music in the evolution, development, and operation of the human brain. Although it sometimes
runs off on self-indulgent tangents (such as the chapter about meeting Francis Crick), the core of the book
revolves around musics signicance to humans, why we are programmed to have a disposition towards
music (and certain musical styles over others), as well as what use music serves in other brain functions.
2 Main Point of Your Brain on Music
Pinning the main point of the book down to just one thing is rather difcult, as the book seems to me to be
closer to a series of essays involving the brain and music rather than a monolithic whole. I dont consider
this a bad thing, as the book doesnt seem to try to present any unied theory, but rather considers several
facets individually. That being said, the book seems to focus around three main points: music utilizes
a large variety of basic brain functions, it is closely tied to emotion, and seems to be advantageous to
survival in line with Darwinian natural selection.
2.1 Music and Brain Functions
It is sometimes claimed that swimming is the best exercise one can do, since it requires one to work nearly
every group of muscles. Music can be thought of as the brains analogue to swimming. In its most basic
and passive form, it exercises timing functions, matches patterns and makes predictions. Pay a little closer
attention, and spatial details, such as the size of the room, become clear. Perhaps youre listening to Jimmy
Page of Led Zeppelin play guitar with a violin bow, Thurston Moore of Sonic Youth shoving a screwdriver
into his guitar, or Kevin Shields of My Bloody Valentine shoegazing a wall of sound out of a single guitar,
and simply pondering over how such strange sounds could be produced. At the more active end, you may
be engaged in creative pursuits, performing, composing, or engineering music.
It should be fairly obvious that even simply listening to music hones a variety of brain functions.
Looking back to Jeff Hawkins book, On Intelligence, one is immediately reminded of the themes of
prediction, pattern matching, and invariant representations. Despite the fact that a song can vary greatly
among live versions, covers, etc., it is not difcult to recognize any of these variations of a song we already
know. Likewise, we expect certain patterns to hold, such as time signature or overall style, and can be
pleasantly surprised when artists slightly violate these expectations.
Overall, one of the dening factors of music enjoyment is this twist in expectations. Too little variation
is seen as boring, while too erratic of a composition sounds incomprehensible. Thus, little children who
1
have no real music experience tend to enjoy very simple music, as just noticing the pattern is difcult
enough. Casual listening adults expect some variation, and tend to enjoy the semi-stable music of Sinatra,
U2, and Metallica, but are turned off by the chaos of The Velvet Underground or Captain Beefheart and
his Magic Band.
The complexity of music is described by the degree to which expectations are broken and also the dif-
culty of noticing a pattern out of seeming chaos. Variation can be manifested in a variety of independent
parameters, such as tempo, timbre, and pitch. Although I dont remember it being mentioned in the book,
I would assume that the brain regions associated with operations in these domains begin developing rather
independently, and different experiences result in unequal development of these regions, so a large deal
of musics enjoyment could come from chaos in one domain rmly grounded in the predictable in other
domains.
2.2 Ties to Emotion
Its no secret that music, as a form of art, attempts to express that which is difcult to express adequately
in words. Music in movies sets the tone (fast-paced music during a chase scene), prepares us for what is
coming next (the approaching shark song in Jaws), or misleads us (any horror lm where suspense is built
out of nothing). Ballads try to go beyond simple statements of love, while religious hymns express many
of mankinds deepest hopes and desires.
All of these are closely tied to emotion, and it is generally accepted that any emotions and our perceived
signicance of an event correspond to one another. Neuroscientists point to the relation of the amygdala
and memory, as well as the evolutionary need for quick responses to dangerous events. When music can
easily be coupled with emotions, it is easy to understand the widespread signicance of music across
human cultures.
This also leads to the interesting effect that as to why we tend to enjoy the music we heard as teenagers.
The teenage years are an emotional roller coaster, so it is no surprise that music is one of the details which
become well-preserved in memory from that period.
It does amaze me how well the two are linked, though. For example, I remember driving several years
ago to see my then-girlfriend on a particularly nice day, with a Neutral Milk Hotel album in my car, and
just thinking how happy I was at that time, and now every time I hear the opening line of the third track
on that album, I not only immediately think of this, but I can remember very specic details, such as the
spot on the road where my car was, accurate within a few meters.
2.3 Natural Selection
Since music appears in every human culture, one must gather that it was either advantageous for survival,
a byproduct of something advantageous, or universally accepted as a desirable quality in a mate. Letvin
seems to lean on the rst, but also implies that it may be a combination of all of these factors.
Asmarter creature is more likely to survive than a dumber, but otherwise equal, counterpart, and Letvin
establishes earlier that music tends to involve large parts of the brain. Thus, applying music as a form of
mental exercise, it would make sense that those creatures practicing music would more effectively develop
their brains, increasing the odds of survival. Music can also serve for communication purposes, and simple
tone patterns may be good substitutes for basic communication, such as yes/no or group hunting orders,
so music would be just as necessary to survival as language.
Other scientists claim that music is a spandrel, a byproduct of something useful. This probably has
some degree of truth in it. After all, it is not difcult to think of situations in which good recognition
2
of timing, pitch, and timbre would be useful to most any animal. Thus, music may simply play on our
necessary natural capacities in such an enjoyable fashion that it managed to easily arise and stick around.
Sexual selection may also be related. Supposing music to be a completely useless skill, the mere
means to develop such a skill would be indicative of a creature with its more pressing affairs, such as food
and shelter, already so well taken care of that it can squander its efforts elsewhere. In short, it may be
equivalent to the peacock tail or the sports car.
Im no evolutionary biologist, but Id say that all three have something to contribute. I can think of
situations in which music would be directly advantageous and of important survival skills indirectly related
to music. Meanwhile, its no secret that successful musicians can sleep with pretty much whomever they
want, so it certainly must play a role in innate preferences in mate search.
3 Application to the Construction of Intelligent Machines
This book struck me as a good extension to On Intelligence. Like the Hawkins book, Brain on Music was
teeming with ideas related to pattern recognition and prediction. It mainly served to reinforce my faith in
Hawkins ideas.
One thing the music book focused a bit more on is the role of emotion. Im not exactly sure how
one would make some sort of emotion processing unit articially, but in the human brain, it is critical for
sorting the useful from the irrelevant. Perhaps we could build a system very different concept of emotions
than could serve as an analogue to human emotions; that might prevent an intelligent system to get caught
up in superstition and imam. On the other hand, many of the most important modern conicts, such
as Israel/Palestine, are incredibly emotionally charged, and I cant see an intelligent computer coming up
with good solutions to these problems without being able to properly understand the emotional stances of
the involved parties.
Towards the end of the book, there is a brief comment in passing about mirror neurons, which re both
when observing an action and when performing the action, and are important for learning. Supposing we
can get a brain-like architecture successfully prototyped in hardware, the next biggest challenge will be
the training of such a system. These mirror neurons seem like a good place to look for inspiration.
4 Concluding Thoughts
I found the book to be well thought out and a pleasant enough read, but Im not entirely sold on the amount
of content it held which was relevant to the course. Overall, it seemed much of the important content was
a restatement of the Hawkins book. I listen to a wide variety of music, play a few instruments, can read
sheet music, and know basic music theory, so much of the material was familiar, but I imagine to someone
who is not, some of the content was difcult and not particularly important.
Having started on the nal book for this class, Ive noticed that there isnt a book which focuses
on superstition, trickery, or fallacious thinking, but that seems like a highly relevant complement to the
Hawkins book. On the topic, Ive only read Michael Shermers Why People Believe Weird Things, but there
are enough books on the topic that I imagine one would be well-suited for this class. Perhaps something
from that eld would be a better choice than the music book.
3
This is your Brain on Music
Thomas Knauer
Throughout the book Levitin is contrasting a prototype theory with an exemplar
theory for storing memories. The first concept holds that only an abstract invariant
version of the perceived data input is stored. Whereas the second one implies that
everything is explicitly saved in a conceptual memory system. Although I personally
felt as Levitin opposes the detailed record tape version, at some point he starts
doubting. Further he supports his doubts with some examples which show that one is
able to recall tiny little details from their past. How is it possible to remember certain
details with a simple invariant pattern? He finally holds that:
we are storing both the abstract and the specific information contained in
melodies. This may be the case for all kinds of sensory stimuli.
As I was convinced by Hawkins idea of using invariant representations
retained through patterns in our brain, I stopped thinking about other possible storing
mechanisms. But there was one thing Hawkins did not mention in his book, namely
how we can recall stored things. Levitin, on the other hand, is promoting the multiple-
trace memory model as a possible solution. It assumes that:
context is encoded along with memory traces, the music that you have
listened to at various times in your life is cross-coded with the events of those
times. That is, the music is linked to events of the time, and those events are
linked to the music.
In theory everything we observe is potentially stored and we only need the
right cue to recall it. In the authors opinion memory is encoded in groups of neurons
and by recreating the required configuration it can be retrieved. Here is where music
comes into play, respectively musical pieces. They can give crucial hints in order to
remember certain things. Songs, among other sensory stimuli, can retrieve almost
forgotten memories.
Let us continue with the mind game that theoretically everything can be
remembered by getting the right cue. The ability to remember things then could be
maximised by having a unique hint for everything saved on your personal mind hard
disk. The obvious problem here is that we have only a limited amount of cues. What I
want to point out is that if there are several clues available for one and the same
memory cell it becomes easier to evoke it. In addition, along one memory trace other
traces can arise and reveal whole memory networks. For this reason thinking about
your 10
th
birthday might remind you of details from your childhood that you had not
thought of for a long time. Associating one and the same hint with too many
memories, however, can lead to the opposite effect. That is why you cannot use a
popular song, which you have heard again and again in different situations, as a cue;
since there are too many memory traces having the same source.
To sum up I experienced the discussed topic as being the most interesting
one. It was complementing ideas of Jeff Hawkins and brought up some new ideas.
However, I still do not feel well informed of how one explicitly recalls memories and
how the theatre of minds can be imagined.
Using the knowledge of the book This is Your Brain on Music to build
intelligent machines might sound unconnected at the beginning. Interestingly, these
research areas have more in common than one might expect.
What is an intelligent device? How do you define intelligent behaviour? And is
an intelligent machine supposed to be inerrable? General speaking artificially created
intelligence must work properly under all circumstances. People tend to lose their
trust in machines if they feel overburdened by their uncontrolled behaviour. Now think
in terms of intelligence as an imitation of the human brain, mind and cognition.
Incorrect decisions made by humans are simply handled as human failure. Yet hardly
anyone mistrusts human confidence. Is that not contradictory? Copying our
brainpower but not allowing wrongdoing? In my opinion it is essential to find a trade-
off between acceptable errors and degrees of intelligence.

Furthermore, I personally attach importance to the mentioned trust issue.
While I was working as a student trainee at the Institute for Data Processing my
supervisor was doing research in the field of Trust between Cognitive Vehicles. This
provided me with some insight information in this area and taught me how difficult it is
to define trust in an abstract way. Once a sophisticated framework is established the
next challenge is to teach how trust, reputation and other factors can be combined
and adjusted by participants of a trust network. One main approach was to find out
how fast wrong information spreads out within a mesh for different trust
implementations.
1
This leads to the origin of trust, namely security. Sometimes it
feels like while rushing from one achieved success to another we forget about the
new created potential for hackers and other bad guys to manipulate and misuse
intelligent devices. Thus I think that more attention should be paid to this topic as its
importance will even increase in the future.

A representative for trust in combination with intelligent devices is how to
decide which source of information is most reliable if there are multiple sources
available. Another concrete example is to consider trustfulness in terms of the
possibility to influence or obscure either the device itself, the way of data collection by
sensors, or, as mentioned before, the data origin.

Much of what we perceive contains missing and obscured information. Thus it
is necessary to implicate the environment when creating intelligent procedures. One
and the same intelligent device probably performs different outside than inside which
can lead to unexpected behaviour. In fact, miscellaneous surroundings affect the

1
http://mediatum.ub.tum.de/doc/997457/997457.pdf
data we are receiving with our set of sensory organs. Since our brain is steadily
adapting to the given environment it seems that our sensation is working somewhat
independent to it. On that account, for example, we usually do not need to
understand every single word in order to understand a sentence. For working out the
message two steps are required. The first one is to readjust the organs to the
ambience and the second one is to fill in the occurring gaps. How we can use this for
our purposes is discussed in the following.

I would like to introduce one approach which is performing with a perceptual
system that can restore information. The basic idea is to create a multi sensory
perception system with a parallel working feature extraction. The features are
calculated independently and the outputs are continuously updated. The main task of
a multidimensional framework is to support the detection of the desired information.
The following paragraph will illustrate the discussed concept.

Think about creating a device for detecting fire and setting of a fire alarm. The
most common concepts are using smoke detectors to become aware of an
emergency. Now this system can be expanded to a more sophisticated multi sensory
arrangement by adding an infrared camera and a microphone. The IR camera can be
used to identify fire pockets before enough smoke is produced in order to activate a
smoke detector. The sound pick-up can be used to extract fire specific sounds. All of
them are working independently. An intelligent device can be trained to react properly
in different situations as it comprises and controls all of the extracted information.
With this concept it would be possible to distinguish between a case of emergency
and a simple candle light. Further it would work faster and more reliable than present
systems and could help to reduce false alarms and would thus provide more safety.
The predicting aspect for this system is that, for instance, the sound of an explosion
or a new unusual source of heat could be compared to previous happenings or
predefined patterns and with its result corresponding actions can be operated.

Levitin holds that it is a tricky issue to categorise things with a specific scheme.
He tries to define characteristics of a typical bird; having wings, being able to fly and
being monogamous. Further he is asking whether or not a penguin fits into this
definition. It does in some ways, but it is very far from the stereotype. Indeed, there
are approximately 8,800 species of birds. With this and some other examples he
points out that it is not always possible to draw a strict line between categories in
order to satisfy all cases. Most of the time, however, it is not necessary to take all
individual cases into consideration. A limited number of possible input data can
reduce the possible outputs. With this knowledge, instead of using a simple Fast and
Frugal Tree approach, it could be more accurate to use an overlapping and modified
prototype theory where one object can be part of several categories, as it was
proposed by Lakoff
2
.

2
http://www.tu-chemnitz.de/phil/leo/rahmen.php?seite=r_wiss/steffen_hecke.php
!"#$ &$ '()* +*,#- (- .)$#/
!"# %&'#(&# )* + ",-+( ).%#%%')( /0'& 1+-.#0%

At one point of ieauing this book - I askeu myself why is Ni Levitin even
seaiching foi answeis. Is he tiying to cieate music that will affect people
psychologically, oi he is he just innately cuiious. I think the lattei. Whilst
he uoes not aim to map the biain anu what happens wheie, he is moie
conceineu with the how anu why things happeneu in the biain as a uiiect
iesult fiom listening to music. Nany of the iueas that he iefeis to, oi
stuuies that he has peisonally conuucteu, aie quite inteiesteu. With
iespect to the biain, he talks about how the biain is uiffeient in people
who aie accomplisheu musicians compaieu to the aveiage laypeison. I
foiget which pait of the biain is enlaigeu, but he iefeis to a pait of the
biain that can giow to be laigei thiough piactice, anu that musically
inclineu people uo actually have uiffeient biains to tone ueaf kaiaoke
singeis like myself!

To be ciitical at the same time the book uiags in with a seemingly infinite
numbei of explanations anu analogies that hau me losing inteiest veiy
iapiuly in what he hau to say. Bowevei his iueas weie quite stimulating,
once again the topic of invaiiant iepiesentations is funuamental to
listening to music, anu how we can so easily iemembei a song. I finu this
aspect of music anu oui biains to be the most inteiesting concept of the
book. I myself have pioblems iemembeiing names of songs (oi even
people foi that mattei), but I can iemembei tunes anu iecognise tunes
(anu people similaily) with ease. I can iemembei basic choiu
piogiessions, when the pitch shoulu inciease oi ueciease, anu many othei
factois. The way in which music is stoieu in my biain, as invaiiant
iepiesentations is quite accuiate. I can iecognise a familiai song whethei
it is playeu on a uiffeient instiument, with a uiffeient pitch, louuness,
tempo, even ihythm. The biain is so well auapteu to making these
connections that oui intelligence goes fai beyonu that of computeis.

The case of labelling mislabelleu songs on people's NPS uatabases was
simple foi a computei to iectify - as easy as scanning a baicoue at a
supeimaiket. Bowevei this is fai fiom intelligence. Computeis stiuggle to
iuentify music if it is slightly uiffeient, in any way. Be it a covei banu
(which is highly iecognisable) oi simply a faulty set of speakeis that uo not
iepiouuce hi-fi sounus, computeis aie in no way at the same level as
humans foi listening to anu uistinguishing sounus.

Fiom ieauing this book, music has become something so much moie than
pleasing sounus. The psychological piocesses involveu aie not simple
piocesses, such as sounu waves ieaching the eai > conveiteu to electiical
signals > simple emotional iesponse. Biffeient pitches fiie in youi biain at
specific fiequencies, uiffeient piocesses occui foi eveiy song. Nusic is
something amazing that has been evolutionaiily uevelopeu fiom (I woulu
now like to agiee with Baiwin) mating ceiemonies in piimitive human
populations. Nusic is useu in Austialian Aboiiginal tiibes to tell theii
histoiy, the 0lu Testament useu to be oiateu to hymn music - music also
helps to iemembei. It is not a meie 'paiasite' of communication, foi I think
it is too inteitwineu with communication to call it a paiasite - they aie
almost one anu the same. Nusic communicates emotions, anu in that way
it is a meuium foi stoiy telling. Listen to song lyiics that you heai on the
iauio - it is a stoiy. Shitty teen music is an exception, with teiiible music
such as Rebecca Black's "Fiiuay" (lyiics go along the lines of, "shoulu I sit
in the fiont seat oi the back seat", as if it is some soit of moial uilemma - it
is a tiuly painful song to listen to). Whilst Levitin mainly focuses on
listening to music in iegaius to emotional ieactions, I think playing of
music is also an outlet foi many people. A lot of veiy populai music stais
have consequently committeu suiciue - anu my peisonal explanation may
be a bit off line - but it is my opinion. I think that some musicians play anu
compose music to fiee themselves fiom negative emotions, anu sometimes
that can be too much foi them, leauing to uiug anu alcohol abuse, along
with othei pioblems. As such I think that music is a highly psycho-
emotional concept that can be pleasing to most people, eithei playing
music oi listening to it.

The emotional attachment to music is also quite inteiesting to me. I can
peisonally think of seveial songs that will always biing memoiies floouing
back into my minu. 0ne of Biuno Nais' songs, about catching a gienaue foi
lovei, will foievei ieminu me of uiiving aiounu Biitish Columbia in
Canaua. It was foievei oveiplayeu on the iauio, anu will always ieminu me
of the positive memoiies of ciuising aiounu on ioau tiips with mates to go
skiing. I am suie that I will also have songs fiom my time heie in ueimany,
such as Fliegeilieu, anu Living Next uooi to Alice - fiom 0ktobeifest of
couise! Without music in those instances, my memoiies woulu be so much
less than what they aie foi me mentally.

So what can be taken fiom Baniel's ieseaich anu to be useu in ieality, to
somehow bettei oui society, anu how people act anu feel. I finu his iuea
about peisonal iauio stations to be quite inteiesting. Be mentions towaius
the enu of the book that in the neai futuie peisonal iauio stations may be a
ieality. A station that is fine tuneu anu manicuieu ovei time to ensuie that
it plays music accoiuing to an algoiithm of one's peisonal music tastes,
theii openness to new styles of music - all leauing to what coulu be teimeu
a 'peifect' iauio station. What goes beyonu this notion of a peisonal iauio
station is that theie is an element of contiol that can be extiacteu fiom the
algoiithm that selects music. Since Levitin has pioveu thiough eviuence in
his novel, uiffeient kinus of music anu sounus make uiffeient paits of the
biain fiie, anu encouiage uiffeient hoimones to be ieleaseu in the biain.
Bopamine, foi example, to ielax people, has been founu to be ieleaseu in
the biain when ceitain music is listeneu to. 0n a siue note, even as I wiite
this essay, I listen to classical music, as I believe it calms anu helps to
geneiate claiity, especially in times of mental wiiting blanks!

Levitin explains that the emotions we expeiience fiom listening to music
aiise fiom the pieuiction mouels insiue oui biain, that happen
subconsciously. Be says that if a song is too pieuictable, it in effect 'ovei
satisfies', anu pioviues little stimulation. But when a song iequiies moie
complex pieuiction mouels within the biain, when expectations aie met,
the biain iewaius itself with emission of ceitain hoimones. Be even says
that when composeis go against these pieuictions, oui biain can
sometimes also enjoy these vaiiances as a way foi the biain to 'woik
haiuei' to accuiately pieuict the next choiu piogiession.

This automateu intelligence coulu be quite an impoitant piogiam to be
uevelopeu in the neai futuie. Peisonally, I uo not take the time to
uownloaubuy music that I like anu to install it upon my iPou, categoiise it
into playlists anu favouiites - all on top of having to caiiy an iPou aiounu
with heauphones. I finu it too much of a nuance, anu I uo piefei just to
listen to a iauio station that has an infinite numbei of uiffeient songs, as
well as the iauio stations that aie known to always play those classic hits! I
think that is it not beyonu the iealm of computei intelligence to eventually
aiiive at a point in time when it is able to simulate music accoiuing to
piefeiences in ihythm, tempo, gioove, beat - many uiffeient factois. As I
have commenteu in my pievious iambling essays, I think that theie is a
fine line that shoulu not be ciosseu by computei intelligence that. I think
that human input is still iequiieu anu shoulu not be eliminateu fiom music
especially!

The human component of music is quite integial - the emotions conveyeu
by a human thiough singing, oi playing music can senu cleai messages to
the listenei. Levitin alluues to this when he mentions Fiank Sinatia, anu
how all his music post 198u was sung with the "satisfaction of someone
who just hau someone killeu" oi along those lines. Bowevei pie 198u he
sang with passion anu stiaineu eveiy note in a specific way in oiuei to
cieate uiffeient emotions.
!" $%&'(%)* &+ ,$%-* .* /&'0 102-+ 3+ !'*-4,
8y Marlus Loch
.+)0&
uanlel !. LevlLln sLarLed hls career as record producer and sound englneer whlle belng a muslclan
hlmself. Pe laLer Lurned lnLo a neurosclenLlsL speclallzlng ln muslc percepLlon and cognlLlon. ln
hls book 1hls ls ?our 8raln on Muslc: 1he Sclence of a Puman Cbsesslon(2006) he comblnes
Lhe flelds muslc and neurosclence Lo glve an overvlew Lo Lhe layperson, explorlng Lhe connecLlon
beLween Lhe sub[ecLs. AfLer an lnLroducLlon Lo some muslcal baslcs (whlch reanlmaLed many
forgoLLen memorles from my hlgh school muslc classes) he Lalks abouL why we llke muslc and
whaL we llke ln muslc (much ls abouL fulfllllng and vlolaLlng expecLaLlons) and why we llke whaL
we llke (we mosLly acqulre our muslcal LasLe ln our chlldhood). ln Lhe laLer chapLers he dlscusses
muslcal experLlse, always sLresslng Lhe polnL LhaL everyone ls somehow an experL ln muslc as
Lhere are dlfferenL klnds of experLlse. 1he lasL chapLer ls abouL where muslc comes from from an
evoluLlonary polnL of vlew.
56'-7 2+7 40"*)266-897 -+)966-(9+49
AnoLher memory from my Llme ln hlgh school was evoked when LevlLln wrlLes abouL
caLegorlzaLlon and how our muslc memory works. ln my phllosophy class (ln Lhe book cognlLlve
psychology ls referred Lo as emplrlcal phllosophy") ln 11Lh grade we had a guesL speaker: rof.
C. Curlo, a neurosclence professor from CharlLe 8erlln. Pe gave a class abouL freedom of wlll and
wlLhln Lalked abouL Lhe dlfference ln lnLelllgence beLween young and older people: he clalmed
younger people have a fluld lnLelllgence", enabllng Lhem Lo qulckly plck up new skllls and solve
problems. AdulLs on Lhe conLrary have a crysLalllzed lnLelllgence", Lhey have a sLrucLured
framework of knowledge based on experlence and prlor learnlng, lL becomes more dlfflculL Lo
plck up Lhlngs ln a new fleld, buL easler Lo apply exlsLlng knowledge or Lo classlfy ob[ecLs wlLhln
our developed framework. l remember how l llked Lhe ldea how we developed a sLrucLure of our
world and sorL new lnformaLlon lnLo LhaL sLrucLure.
1hls correlaLes Lo much from LevlLln's book. ln Lhe chapLer AnLlclpaLlon" he Lells us how we
bulld our frameworks: An lmporLanL way LhaL our braln deals wlLh sLandard slLuaLlons ls LhaL lL
exLracLs Lhose elemenLs LhaL are common Lo mulLlple slLuaLlons and creaLes a framework wlLhln
whlch Lo place Lhem, Lhls framework ls called a schema" (p.113). Pe clalms we have muslcal
schemas as well for all klnd of Lhlngs: sLyles and eras, cerLaln bands, arLlsL and genres, coverlng
rhyLhms, chords, Lyplcal moLlfs and sequence, eLc. - we learn whaL Lhe legal moves ln Lhe muslc
of our culLure" are.

1he auLhor also descrlbes how Lhese schemas are developed mosLly ln our youLh: we sLarL
already prenaLal, ln Lhe womb of our moLhers. 8y Lhe age of Lwo chlldren sLarL Lo show
preference Lowards Lhe muslc of Lhelr culLure (p. 230), Lhen Lhe age Len Lo eleven becomes a
Lurnlng polnL when all chlldren sLarL Lo care abouL muslc, even Lhose who dldn'L care so far (p.
231). 1he mosL lnfluenLlal perlod ls Lhe age around fourLeen when Leenagers sLarL Lhelr
emoLlonal self-dlscovery" and use muslc as soclal ldenLlflcaLlon (p. 231). 8y Lhe age of LwenLy
Lhe braln ls mosLly flnlshed" and lL becomes seemlngly more dlfflculL Lo learn new skllls (p. 233)
- also our framework for muslc ls mosLly seL.

1hls framework (and Lhe conLalnlng schemas) descrlbe how we Lhlnk muslc has Lo be and whaL
lLs legal moves" are - Lhese are our expecLaLlons Lowards muslc. As Lhe auLhor goes on and
polnLs ouL several Llmes ln Lhe book, we llke muslc LhaL fulfllls our expecLaLlons, buL vlolaLes
Lhem every now and Lhen:

!"#$ &#'())*+ ,#())*+ -./ &$-'* 0$ $12$'($.,$ 3'45 56*(, -'$ &#$ '$*6)& 43 #-7(.8 46' $12$,&-&(4.*
-'&36))9 5-.(26)-&$/ :9 - *;())$/ ,4524*$' -./ &#$ 56*(,(-.* 0#4 (.&$'2'$& &#-& 56*(,< =>? "#$
*$&&(.8 62 -./ &#$. 5-.(26)-&(.8 43 $12$,&-&(4.* (* &#$ #$-'& 43 56*(,<@

lf lL ls Loo slmple we percelve lL as borlng, lf lL ls Loo complex we can'L flnd any sLrucLure or
famlllarlLy and don'L llke lL (p. 233). 1he Lerms slmple and complex of course depend on our
framework and how advanced lL ls ln a parLlcular sLyle or genre.
1hls correlaLes well Lo rof. Curlo's descrlpLlon of fluld and crysLalllzed lnLelllgence. ln our
chlldhood we use our fluld lnLelllgence and plck up Lhe rules of our culLures muslc. 1he muslc we
llsLen Lo ln LhaL Llme shape our undersLandlng of whaL muslc ls and whaL Lo expecL of lL. 8y Lhe
Llme we become adulLs Lhese expecLaLlons represenL our LasLe ln muslc. We don'L learn" any
more new muslc (unless we acLlvely sLrlve Lo do so), buL we apply our ,'9*&-))(A$/ (.&$))(8$.,$.
We llsLen Lo muslc and Lry Lo flL lL lnLo our framework and based on LhaL declde wheLher we llke
lL or noL.
!9:&0";<097-4)-&+ =02:9>&0? >-)% :'6)-<69;)0249 :9:&0"
ln chapLer flve where memory and caLegorlzaLlon are dlscussed Lhe auLhor descrlbes one
problem: lL ls easy for a compuLer Lo recognlze" a song by look up. 8uL lL ls very dlfflculL Lo
recognlze Lwo dlfferenL verslons of Lhe same song by comparlson. 1he braln does Lhls wlLh
ease, buL no one has lnvenLed a compuLer LhaL can even begln Lo do Lhls." (p. 133) he
lmmedlaLely glves a reason ln Lhe nexL senLence: "1hls dlfferenL ablllLy of compuLers and
humans ls relaLed Lo a debaLe abouL Lhe naLure and funcLlon of memory ln humans."
As scholars of !eff Pawklns Cn lnLelllgence" Lhls rlngs a bell of course, memory as a cruclal parL
of an lnLelllgenL sysLem. 1here are more parallels: Pawkln's lnLelllgenL sysLem as memory-based
predlcLlon sysLem corresponds perfecLly Lo Lhe human's percepLlon of muslc. 8ulld a frame work
ln memory by paLLern recognlLlon, predlcL whaL comes nexL and be alerL (or ln Lhe case of muslc
pleasurably surprlsed) lf Lhe predlcLlon wasn'L qulLe correcL.
When we Lhen read LevlLln's descrlpLlon of Lune recognlLlon Lhe parallel ls compleLe: 1une
recognlLlon lnvolves a number of complex neural compuLaLlons (.&$'-,&(.8 0(&# 5$54'9. lL
requlres LhaL our bralns lgnore cerLaln feaLures whlle we focus only on feaLures LhaL are
lnvarlanL from one llsLenlng Lo Lhe nexL-and ln Lhls way, exLracL (.7-'(-.& 2'42$'&($* of a song."
(p 133). 1he absLracLlon of lnvarlanL feaLures ls also a cenLral Lheme ln Pawkln's book.
ln oLher words LevlLln's descrlpLlon of how we percelve muslc reads llke a manual Lo Pawklns
lnLelllgenL machlnes. So explorlng Lhe debaLe abouL Lhe naLure and funcLlon of memory ln
humans" (see above) seems llke a good polnL Lo help wlLh bulldlng lnLelllgenL sysLems.

ln Lhls dlscusslon Lwo vlews are lnLroduced. Cne ls Lhe ,4.*&'6,&(7(*& &#$4'9 (relaLlonal memory),
clalmlng Lhe braln only sLores relaLlons beLween ob[ecLs and ldeas, buL noL necessarlly deLalls
abouL Lhe ob[ecLs Lhemselves" (p. 133). 1hls lmplles we acLually ,4.*&'6,& our memorles every
Llme we access Lhem and deLalls are more of a sophlsLlcaLed guess.
1he second vlew ls Lhe '$,4'/B;$$2(.8 &#$4'9 (absoluLe memory), clalmlng memory sLores mosL
our experlence accuraLely and ln deLall.
unforLunaLely Lhere ls a loL of evldence for boLh Lheorles. An easy and lnLulLlve example: every
one of us wlll recognlze a song we know, even lf lL was Lransposed by some Lones. We wlll also
recognlze lL, when lL ls played fasLer or slower Lhan Lhe orlglnal. 1hls would supporL Lhe
consLrucLlvlsL Lheory, as Lhe dlfferenL verslons are ldenLlfled by Lhe relaLlonshlp of Lhe sounds.
Cn Lhe oLher hand everyone wlll noLlce, when a song ls played dlfferenL Lo a famous
performance llke a radlo verslon llsLened Lo all summer long. ln facL mosL people wlll noL only be
able Lo Lell Lhe dlfference, buL be able Lo reproduce Lhe famous verslon (as far as Lhelr muslcal
means wlll allow). 1hls supporLs Lhe record-keeplng Lheory, as lL requlres absoluLe lnformaLlon
abouL a cerLaln performance.

Cne lnLeresLlng aspecL LhaL comes up ln Lhe dlscusslon abouL Lhe Lwo memory sysLems:
apparenLly we rely on sequences Lo memorlze muslc. MosL humans and even experL muslclans
can'L evoke Lhe memory of an arblLrary momenL ln a song. lf we wanL Lo remember a speclflc
parL, we have Lo scan Lhe song from Lhe beglnnlng or anoLher slgnlflcanL landmark wlLhln Lhe
song. LevlLln concludes: 1hls suggesLs LhaL our memory for muslc lnvolves hlerarchlcal
encodlng-noL all words are equally sallenL, and noL all parLs of a muslcal phrase hold equal
sLaLus. We have cerLaln enLry polnLs and exlL polnLs LhaL correspond Lo speclflc phrases ln Lhe
muslc [.]". AnoLher Lwo lmporLanL keywords from Pawklns book: sequences for memorlzaLlon
and applylng a hlerarchy.

As a soluLlon Lo Lhe Lwo confllcLlng memory sysLems Lhe auLhor lnLroduces a hybrld Lheory: Lhe
mulLlple-Lrace memory. Accordlng Lo Lhls Lheory we sLore speclflc (absoluLe) lnformaLlon as
record-keeplng would suggesL. lurLhermore lL ls suggesLed: As we aLLend Lo a melody, we musL
be performlng calculaLlons on lL, ln addlLlon Lo reglsLerlng Lhe absoluLe values, [.] we musL also
be calculaLlng melodlc lnLervals and Lempo-free rhyLhmlc lnformaLlon [.] creaLlng a plLch-free
LemplaLe [.] ln order Lo recognlze songs ln LransposlLlon. [.] [1hls] suggesL[s] LhaL we are
sLorlng boLh Lhe absLracL and Lhe speclflc lnformaLlon conLalned ln melodles."
As Lhls Lheory clalms Lo preserve Lhe conLexL of a memory, lL would explaln how we llnk muslc Lo
cerLaln slLuaLlons ln our llve and how llsLenlng Lo a song can con[ure Lhe memory of a forgoLLen
evenL far ln Lhe pasL.

So we saw how closely llnked Lhe percepLlon of muslc ln Lhe sense of uanlel LevlLln ls Lo
lnLelllgenL machlnes ln Lhe sense of !eff Pawklns. A key feaLure of boLh seems Lo be Lhe human
memory sysLem. 1herefore LevlLln lnLroduced Lhe mulLlple-Lrace memory Lheory somewhaL
comblnlng Lhe record-keeplng and Lhe consLrucLlvlsL approach by sLorlng absoluLe, buL also on-
Lhe-spoL calculaLed (absLracLed) lnformaLlon. Maybe a furLher developmenL and analysls of Lhls
memory sysLem could - applled ln compuLer sysLems - dellver a breakLhrough for lnLelllgenL
machlnes.
This Is Your Brain on Music
The Science of a Human Obsession


Alexandra Marinescu
Brain, Mind and Cognition



We are [...] under the illusion that we simply open our eyes and we see. A
bird chirps outside the window and we instantly hear. Sensory perception
creates mental images in our minds representations of the world outside
our heads so quickly and seamlessly that it seems there is nothing to it.
This is an illusion. [...] Our perceptual system is supposed to distort the
world we see and hear. We naturally assume that the world is just as we
perceive it. But experiments have forced us to confront the reality that this
is not the case.

Normally, while reading a book, I have to underline specific passages of
text, which I find of great importance, passages that somehow say and
mean more to me than the rest of the book, passages that manage to
change my beliefs or the way I understand my surroundings, the
environment I live in. The quote above did more than that. I had to write it
down, I had to think about it. Underlining it wasnt enough anymore.

So the world isnt what we perceive it to be. Everything we see or hear or
touch or even taste is just a mental projection of the real world. I found
myself thinking about a normal day in my life starting with the moment I
wake up in the morning and start seeing, perceiving my surroundings.
And I surprisingly got stuck at this point, since my first thought was about
the daylight. If what we perceive as light is only a mental image, the
response of our brain to the sensory perception of an oscillation, can it be
that there is no light at all?

It seems that the human brain assigns a label to everything he processes,
based on the different categories he learns over time. And if we also
recall the beliefs of E. Rosch, we can also understand why something can
be more or less a category member; rather than being all or none as
Aristotle has believed, there are shades of membership, degrees of fit to a
category, and subtle shadings. Our sensors perceive light and we think
of it as a day, or dark and we know that the night has come. We
perceive something in between the two categories, neither a day nor a
night, so we it must be either a dawn or a sunset. But then again, these
are only mental images. Light, as we perceive it, may indeed not exist.
The day is just the label we gave to a category of sensory data our brain
processes. One might jump to the conclusion that we live in the dark, but
that cannot be true, since dark is also just a label.
I kept reading the text passage again and again and couldnt get to a
conclusion. I tried to find answers in the book, but instead I found only
further questions and text passages to write down. For example this one:
Perhaps the ultimate illusion [...] is the illusion of structure and form. So
not only does our world look, sound and feel differently as we all perceive
it, now it also doesnt have a structure at all! It seems that we make sense
of the world depending only on the environment we grow up in, on the
things we learn over the years, on our experience.

To sum up, we do see light or dark instead of a certain wavelength, or a
bed or a table instead of the individual atoms they are made of. They
might not exist in the real world as we perceive them, but that doesnt
make them, in my opinion, unreal, just different. We see a bunch of
organized, oscillating particles and assign them a label based of the
categories we learned over the time; we are the ones giving a structure
and a sense to our world by interpreting our sensory perceptions.




How can the knowledge obtained from the book be used for
building intelligent machines?


I strongly believe there is still a very long way we have to go before
building truly intelligent machines. I also believe that we will not be able to
do that unless we have an understanding of how the brain really works.
Unless scientists come up with a theory that manages to provide a unitary
answer to the raised questions, no breakthrough will be made.

The book of reference may not make a breakthrough, but it most certainly
introduces an entirely new point of view in studying the brain, by
considering music and its effects.

So what if we learn how to endow robots with emotion from the brains
perception on music? Tempo is a major fact in conveying emotion. Songs
with fast tempos tends to be regarded as happy, and songs with slow
tempos as sad. [...] In order to be moved by music (physically and
emotionally) it helps a great deal to have a readily predictable beat. [...]
Music communicates to us emotionally through systematic violations of
expectations. As music unfolds, the brain constantly updates its estimates
of when new beats will occur, and takes satisfaction in matching a mental
beat with a real-in-the-world one, and takes delight when a skillful
musician violates that expectation in an interesting way.

For a robot isnt, of course, a trivial task to react emotional to the different
violations of their expectations, with these expectations being nothing
else but a probabilistic model based on the training data they were given.
But when interacting with people, this could take a totally new turn.

People do react differently when interacting with machines, compared to
the times when they are dealing with fellow humans, mostly because they
know that such machines dont share the same internal structure
reactions, emotions, thoughts as they do. But if we were to apply on
machines what we learned from music, robots would be able to give rise to
emotions only by violating the expectations of the humans.

By adapting the tempo and the timbre of their voice, robots should be
therefore able to communicate different emotions to the people they are
interacting with.

Of course, this is yet another way of tricking people into thinking that
robots are truly endowed with emotions. It still doesnt represent a step
forward in building truly emotional robots.

Another big constraint is, of course, the mind. For cognitive scientists,
the word mind refers to that part of each of us that embodies our thoughts,
hopes, desires, memories, beliefs and experiences. The brain, on the other
hand, is an organ of the body, a collection of cells and water, chemicals and
blood vessels, that resides in the skull. Activity in the brain gives rise to the
contents of mind. [...] Different programs can run on what is essentially the
same hardware different minds can arise from very similar brains.

I do want to believe that there is more to us than an evolved brain. I refuse
to believe that this feeling the self could be an illusion, just as it
certainly feels as though the earth is standing still, not spinning around on
its axis at a thousand miles per hour. And for this reason I also strongly
believe that robots will remain just hardware; they might get to have the
intelligence of a human brain, but they will never have a mind the self.
Audrey Pedro
This Is Your Brain On Music from Daniel Levitin
12/01/2012

Daniel Levitin is interested in music (he works as a music producer) and in particular
in how music is processed by the human brain in order to understand why is music such a
big deal for most of the people. His book This Is Your Brain On Music presents his theory
and researches results on the topic.
Two points were to me particularly interesting in this book, the first one because of
personal experience and the second because of its power. What I found very impressive in
this book due to personal experience is the theory about how music tastes are created.
Indeed I always found it very funny that people stay so attached to the music they were
listening in their youth and when they are listening to it they become the young person they
were almost instantaneously. This phenomenon can very easily be observed in wedding for
example.
Here Daniel Levitin gives an explanation: as a language or any other learned skill,
music structures are acquired without too much effort in the youth. The parallel made with
language was the one that spoke to me. Young children are able to learn a language without
any grammar lessons and so on: they just ear it and memorize what other say. Then they
build language structures and become native speaker which means they master the
language with no much effort. In comparison to older people learning a new language the
way young children become expert in their mother tongue is astonishing. For music it is
exactly the same: until a certain age learning new structures is very easy and then it
demands some effort. Yet this learning ability decreasing with age has nothing so illogical but
what I found interesting is the combination with the fact that people do like music if they are
familiar to its structure. When I read that I was surprised, it means then that new songs we
like are just similar to other songs we already like, boring no?
Actually not so much because as Levitin explains the same structure allows many
variations (music has many components that can be changed: pitch, rhythm, timbre) and
the known frame is just a safety for the listener. This idea impacted me because of personal
experience (it is even an example used by Levitin about Latin music). I grow up with a
Colombian mother who was listening to salsa, merengue, beguine and other Latin music. I
now appreciate this kind of music very much and I love dance on it but it is very difficult for
me to find dance partners and to put my music on parties. Worst I have the impression that
people dont get the difference between salsa and merengue when I can hear it since the first
notes . Well reading the page concerning that (241 in my edition) was a relief. I finally know
why I feel so alone when listening to Latin music.
I must admit that this book was the first I was not so enthusiastic to read. First of all
the beginning was very technical in music and I think that knowing what pitch, timbre, meter
and loudness are do not help for enjoying music listening. As Levitin says the expert
language is too complicated and creates a separation between musicians and other people
which first was not always there and then is not necessarily justified. Id like to come back to
the parallel with languages: native speakers are able to detect very easily if a sentence is
correct or not but in general they do not know the grammatical rule violated. Concerning
music, people can appreciate music without naming everything. I was hoping that after this
chapter the book would become more interesting but I was disappointed. I did not manage to
get really in the book before chapter 6 and it is now difficult to synthetize ideas about what I
read before this chapter. Chapter 6 gets my attention by presenting what I felt like the core
idea of the book: how music experience is anchored in the reptilian brain and how it calls the
emotion region of the brain (amygdala).
Music is an art from whose medium is sound and silence, definition of music from
Wikipedia mentions the music as an art and it is the most common definition. Art belongs to
the evolved human skills about just creating aesthetic for itself without any vital need
underneath. But in his book Levitin explains that music is something managed by the older
part of the brain, a region we have in common with reptilians such as snakes and that this
region makes the link between music experience and strong emotion reactions. The reptilian
brain allows us to protect ourselves: it normally manages inputs that lead to strong emotion
reactions in order to make us act without thinking too much in dangerous situations. When
we feel burning for example, the reptilian brain will immediately transmit that we have to run
away from fire to save our live and the amygdala is the medium used to make it quick. So
here the theory seems contradictory: how an art can be managed by a region dealing with
vital needs? Levitin explains how music isnt just an art but an evolved skill still being an
instinct (see chapter 9).
The music-as-an-instinct theory is very satisfiable concerning the main question of the
book understanding a human obsession or why is music such a big deal in all societies and
cultures known. It is not just an art but an instinct that we still have and this explains why
mothers sing to their babies to calm them down and why dance is considered as something
so erotic / intimate.
This leads us to the second part of the essay concerning what would be useful in this
book if I was working on an intelligent machine. This book is not so easy to use for building
intelligent machines since the author does not mention any theory about intelligence in
general but focuses on music. Nevertheless understanding music can be considered as a
part of human intelligence.
The first point mentioned above about structures learning and recalling them is an
interesting track. An intelligent machine could learn to build structures in order to apprehend
other unknown things when the structure is similar. This ability could be useful for other
domains of learning (mathematics, linguistic). This characteristic corresponds to a
synthesis ability. It joins the theories of the two previous books that a frame of memorized
known structures that can be adapted to new situations is the basis of intelligence.
This common point emerging from the three first books confirm that intelligence
probably rely on it. Intelligence is the adaptation of known frames to the unkown.


Brain, Mind and Cognition Nadja Peters
Daniel Levitin This Is Your Brain On Music 09-01-2012
Whats your opinion of the most interesting thought?
Where did the book give you inspiration for building intelligent systems and what is your
inspiration?

Why is it sometimes so hard to remember all the interesting things one is supposed to learn in
class? Even though the professors try to teach us in the most passionate way (at least lets
assume that for now), most people have a hard time understanding (e.g. math) and
remembering facts (e.g. differences between programming languages or computer
architectures). On the contrary it is quite easy to remember non-relevant things like the
number of marriages of your favorite actor or the number of goals your soccer team achieved
in the last season. Such facts are usually very interesting, but they do not at all help us passing
the upcoming exams. So why are facts, which we need to learn to move on in our careers, much
harder to remember than unnecessary details about our favorite athletes? As Daniel Levitin
tells in his book This Is Your Brain On Music, we need special cues to retrieve memory from
our brain (Chapter 5, You Know My Name, Look Up The Number). He also tells us that the more
passion is related to a memory, the more likely we will remember it.

So how can we possibly add as much passion to remembering boring things as to things we
are really interested in? Usually the problem is not about the content of a class being boring (at
least this SHOULD not be the problem, although one never finds every subject equally
interesting), but the problem is about sitting down and start learning, memorizing. Although
one has acquired more knowledge and has a great chance to pass exams after doing some
successful learning process, it still involves a lot of stress and pressure. As shown in the
statistics, many people are not able to keep up with that kind of pressure and simply quit their
studies (to keep things correct: being under pressure is of course not the only reason to quit
studies but in these days it is becoming more and more relevant
2
).

.
0
5
10
15
20
25
30
35
40
45
50
Students giving up university
1
Brain, Mind and Cognition Nadja Peters
Daniel Levitin This Is Your Brain On Music 09-01-2012
As this is a topic which is important for every one of us, it is interesting to find ways to improve
this situation and bring more joy into the lives of stressed students. But how are we supposed
to improve this situation? There are different possibilities:

1. Take away the pressure from student
2. Find methods of learning that are so effective that the student feel comfortable
3. Find methods of relaxing that restore the students psychological condition

Possibility one is definitely the easiest one. Taking away the pressure can be done by simply
making the classes less challenging. As that is usually not increasing the esteem of a university
the pressure could also be taken away by increasing the number of semesters that a student
need to achieve a degree.

As we are studying brain, mind and cognition we will not go the easy way but try to figure how
we can use our knowledge to develop solutions for possibilities two and three. So how can we
manipulate the neurons in our brain to make learning more effective and how can we
manipulate them to make ourselves feel comfortable and relaxed (without using drugs or
similar)?

As Daniel Levitin points out, for remembering things we need certain cues. A cue can be
compared to a key to a door. When one possesses a certain key, that person can open a special
door in ones brain and get access to the information lying beyond. Levitin also says that
memories which involve certain emotions can be remembered easier by our brain. How can we
use this knowledge to stimulate our brain? How can we create some sets of mnemonics for
certain topics and fill those with emotions, so we can remember them most effectively? In this
case effectively means that one uses the smallest amount of time possible for learning but can
remember the things for a long period of time. Moreover learning should not be tight to sitting
at the desk and staring at formulas all over again but involve some kind of actions that make it
more comfortable. But how can we achieve that something a person does not like is actually
fun?

To go with Levitin we could pick some music that makes us feel good and listen to it before
learning. Every time we feel bad we can turn it on and get some positive emotions out of it. At
this point, I would like to cite my favorite quote: Repetition, when done skillfully by a master
composer, is emotionally satisfying to our brains, and makes the listening experience as
pleasurable as it is, (Chapter 5, p. 167). But is it really enough to stimulate our neurons with
some really good pieces of music? As we have learned, the brain is of versatile structure. So
what happens if we listen to some pieces of music that we like and then do something else we
do not like? In my opinion the stimulation will lose its effect because the structure of the
neurons will adapt after some time and the association we have with a certain piece of music
will change. We cannot even compare music to drugs although many people like to do this: If
someone takes drugs, that person will simply increase the dose to get satisfied, but if we
increase the dose of music we do not like anymore, it will probably make us feel worse and8
Brain, Mind and Cognition Nadja Peters
Daniel Levitin This Is Your Brain On Music 09-01-2012
frustrated. So in my opinion to keep the positive effect of music we should not use it to make us
happy before doing things we do not like. After some time it will not work anymore.

We found out what probably will not help us learning more effectively, but what is it that WILL?
What we need for a good strategy is not only a good set of cues and some emotions make the
cues more effective but we also need to repeat the things we want to remember all over again.
Repetition is, besides understanding the context, the most important clue to remember things.
But what is able to get stuck in our minds by being repeated all over again and being pleasant
as well? In our everyday lives, most of us are listening to music most of the time. We can
remember a lot of songs and even sing along to a lot of songs? What if we do not sing along to
lyrics sung by Michael Jackson or Elvis Presley but create our own songs, songs we use to
memorize things we have to learn for an exam for example? What if professors would create
songs and record their lessons? So we would not only buy scripts by the beginning of the
semester but also some recordings to listen to on our MP3 players while going to university in
the mornings. After some time we could sing along and remember all the lessons simply by
humming the melody of the song. Sure, this method cannot save us from solving difficult math
equations but in most cases doing practical exercises is not half as boring as memorizing things.
Doing learning sessions could get another meaning if students simply met in a club and started
dancing and singing to the lectures.

As Levitin says in his book, the ability to make music and being affected by music is natural to us
because it is determined by our genes (Chapter 9, The Music Instinct). So when we want to
build intelligent machines, what kind of genes do we need? What are we supposed to tell it so
that it can function properly? Stephen Jay Gould is of the opinion that music is nothing that we
developed to survive, but that it is a by-product by other functions as language. What happens
if our intelligent system developed some by-products that we did not want it to develop?
Whenever we design a system we want to be in control of that system. How can we be sure
that intelligent systems do not develop their own mind? By their own mind I do not mean, that
they decide to take over planet earth because they are of the opinion that human kind is not
able to control the system properly. What if they simply make mistakes? Actually, it would be
normal, because making mistakes is human, but how can we trust in a machine that drives our
car in the wrong direction, for example?

Maybe we could create a machine that records lessons at university and automatically converts
them to songs we can listen to. Those machines would not harm our lives by making mistakes. I
really do think that listening to songs instead of simply sitting at your desk and staring at a book
would improve our learning performance and make the learning process more enjoyable but
who would actually transform the content of a lesson into a song and record it? There is not
only need in people who actually do know the subject but other people who can do the
performance. Although the effort seems quite large on the first glance, I would love to test that
method of learning.
______________________________________________________________________________
1 Das Studentenportal, Studienabbrecherstudie, 2005, http://www.studserv.de/studium/statistik.php (access: 05-01-2012)
2 Nina Zimmermann, Studenten unter Druck, 2008, http://www.spiegel.de/unispiegel/studium/0,1518,569612,00.html (access: 05-01-2012)
Written Discussion on Daniel J.
Levitins This Is Your Brain On
Music
Martin Reverchon
January 9, 2012
1 What is in your opinion the most interesting
thought/idea?
Being only an expert listener and unfortunately having no visible talent in be-
ing an expert performer this book made me joyfully think about how music and
sounds aect us in our daily life. My roommate is a clarinet student at the Uni-
versity of Music and Performing Arts Munich and therefore we tend to talk a lot
about music in general (and I certainly did bombard her with a lot of questions
about the little of music theory we were taught in the book). However, I by myself
never made a connection between music theory and our brain. Therefore, I was
especially surprised by the in hindsight relatively obvious fact that music and
sound like any other perceptional sense we have constantly aect the state our
brain and each individual neuron in it.
Stimulating electrodes deep inside the brain begin to unfold their potential.
Depth-electrodes can cure Cluster-Headaches, muscular spasms or obsessive com-
pulsive disorders. Patients who washed their hands hundreds of times daily can
now again lead a normal life. With the help of depth-electrodes it was even possi-
ble to awaken a patient who was in coma for six years.
I have recently heard about a neuro-technology called controlled reset (CR
R
).
In brain regions that are aected by a certain type of diseases with symptoms of
tremors and tinniti often lack sensory input that inuences the state the aected
set of neurons is in. Without this input it is possible that these cell assemblies
fall into a self-induced state of mutual pathological synchronisation of which
the above mentioned symptoms stem from. Without the input it is impossible to
desynchronize the aicted neuronal areas. This is the point where neuro-implants
1
come into play. An electrode is surgically inserted into the patients brain. Firing
at specic frequencies the electrode is able to weaken and ultimately desynchro-
nize the synchronous ring of the neurons resulting into the relief of the malady.
Surprisingly it was found that in certain cases where a proportion of the visual
or in this case auditory nerve was still intact, no surgical operation is neces-
sary. Presenting the patient a sound pattern similar to the electrodes impulses
resulted into the cure of the symptoms, too! This is another very good example
of how sound aects us. Since all of the above mentioned therapies with depth-
electrodes work, is it possible to aect our brain by mere presentation of audio or
visual patterns to cure these sicknesses? Could it be an even more powerful tool?
Music is able to lift our mood, bring us down, soothe our temper or enrage us.
As shown above it might even cure cerebral illnesses. I think it would be great
if we found the principles in how exactly music and sounds aect our brains. If
we could wield that power like a scalpel surgical non-invasive sound therapies
could be used to cure or relieve a wide variety of mental or structural cerebral ill-
nesses. It could be used to put us in several desired cerebral states. May that be
a soothed state for patients with bipolar disease or excitement and agitation for
lethargic people. It is even thinkable to reach some kind of transhumanism via
sounds and music. We could put ourselves into states where we are able to per-
ceive the world better, where we are able to learn quicker and, on the other hand,
where we can erase unwanted memory (if this is desirable is of course a broad eth-
ical discourse).
2 Where did the book give you inspiration for
building intelligent systems and what is your
inspiration?
I recently had the chance to visit ACE in its lab. ACE stands for Autonomous
City Explorer. The ACEproject envisions to create a robot that will autono-
mously navigate in an unstructured urban environment and nd its way through
interaction with humans. To achieve this, research results from the elds of au-
tonomous navigation, path planning, environment modelling, and human-robot
interaction are combined. However, on rst sight I have to admit that I felt a bit
frightened by its appearance. Besides its bulky built and its massive jaws that
look like they could easily crack a coconut it had a red antenna wiggling in a snake-
like way on its head (which is used to point in dierent direction a pretty use-
less feature as it turned out, I was told by the researchers). Being asked for direc-
tions by this clunker Id rather change the side of the sidewalk than help this odd
2
penguin-shaped massive monstrosity. Designing a more appealing shape for this
robot will certainly wield a higher acceptance during human-robot interaction.
But the right set of sounds could also be employed to make a valuable positive
rst impression, when this robot approaches humans in order to full its objec-
tive. Moreover, during all situations where human-robot interaction is necessary
the right choice of a comforting voice can set a positive mood for human-robot in-
teraction. However, this does admittedly not lead to intelligent machines. It may
just change the way we look at robots and intelligent systems.
Gerd Gigerenzer stated in his book Gut Feelings that a specic intelligent
system I suppose a neural network of some kind was able to learn speech if it
was presented with only simple sentences. I wonder if this was possible for music,
too. Could an intelligent understand music if we presented it with simple chimes
like commercial jingles, simple enough for a toddler to hum. Advancing over a bit
more complex music like minimal music to big symphonies a structure similar to
those that are processing our speech could emerge. Would a similar neural net-
work be able to learn the basics of pitch, rhythm, tempo, contour, timbre, meter,
key, melody and harmony like the dierent cues of speech; grammar, vocabulary,
syntax, semantics and prosody?
Taking Daniel J. Levitins approach of looking at the brain under the inuence
of music could be adapted to analyse the behaviour of neural networks. This would
not only help in the eld of research of neural networks. With the achieved in-
sight in neural networks we could also get more knowledgeable about the impact
of sounds and music on the brain on a functional level.
There are two remaining big questions in this world. The rst one is How
did the universe into existence. The second one is How does our brain work.
Achieving an understanding of the brain on both, the functional and the behavioural
level, would bring us closer to a solution for the latter problem. Thus, applying
music to several dierent aspect of brain research on a biological and a techni-
cal level can be of great interest. This gets even more clear if look at one of Lev-
itins statements in the book: Cosmides and Tooby argue that musics function in
the developing child is to help prepare its mind for a number of complex cognitive
and social activities, exercising the brain so that it will be ready for the demands
placed on it by language and social interaction.. Taking this as an analogy for
the development of technical system, I think that a similar approach could bear
feasible results.
3
Tudor Timisescu

Report on the book This Is Your Brain on
Music


1. What is the main point of the book according to your opinion?

With the book This Is Your Brain on Music, the author, Daniel J. Levitin, intends to familiarize
the reader with basic aspects of musical theory, related to the various aspects of musical
perception and interpretation, with the way all of these aspects are processed by the human
hearing apparatus and the brain and with how they all fit it together to create emotions and
impressions inside the mind.
Music is composed of a large palette of elements, such as pitch, rhythm, tempo, contour,
timbre, loudness, reverberation, meter, key, melody and harmony. All of these elements come
together to create a specific impression on the listener. That we can distinguish between a note
played on a piano and the very same note played on a clarinet or if a piece of music is played in a
small cramped room or a big concert hall is nothing short of miracle, the author believes. Human
perception is far more advanced than any of todays fancy computer algorithms and will be for
much longer.
All of these individual elements are processed in different regions of the brain, from areas
of the neocortex, to the cerebellum (the oldest part of the brain), to the amygdala (the part of the
brain that deals with emotions). This last connection reinforces the profoundly emotional role that
listening or playing music has. The music processing machine follows the layered approach
proposed by Hawkins in his book On Intelligence.
Seeing as how music has a very big emotional charge linked to it, memories that are
associated with musical passages also share this charge. This helps them to be better
remembered, even in extreme situations like those of Alzheimers disease patients.
Music has a natural flow to it. Sounds coming at us without any structure to them will
only be perceived as noise. It is the structure of musical passages that give music its sound and
feel. Anticipation of what will come next in a musical piece is key to what the listener feels.
We have all experienced the feeling that while listening to a particular passage of music we have
never heard before we still kind of know what will come up next. Sometimes though, these
anticipations are violated and a sort of inner turmoil is created. It is through a mix of these two
states that good music is made: if a piece is too predictable, then it becomes boring; a piece that is
too unpredictable sounds foreign. The way composers create these mixes is what separates the
skilled ones from the novices.
Although too simple pieces of music (childrens songs for example) are indeed boring,
complicated music is not necessarily good music. Very often, you will find a great deal of
proficient players, who can play notes at lightning speeds without making any mistakes, but their
playing sounds bland and boring. It is the ones who can transmit the most emotion that are the
most successful (as if pouring their soul into their music) and the human mind is very good at
telling these ones out.
Many studies have been done to determine the impact of nature vs. nurture in musical
ability. There is no clear evidence either way. Some people tend to think that musical talent is
something that only some people possess. They give as examples musically inclined families.
This argument, though, can be countered with the positive reinforcement argument: children that
grow up in such families are more likely receive positive reinforcement for pursuing musical
activities and will be much more inclined to continue to do so. There is also a theory stating that
there is a certain minimum of practice anybody needs in order to become an expert in a certain
field (ten thousand hours). According to this theory, anyone can become an expert as long as he
or she spends this much time honing his or her skills. Such an argument can also be countered by
the fact that different people progress at different rates. As it is with most everything, the truth is
probably somewhere in the middle.
The final chapter of the book is devoted to the evolutionary aspect of music. The
questions on researchers minds are why and how did musical ability develop in humans. The
author challenges one of the major theories of human evolution saying that music is simply a by-
product of other evolutionary traits such as speech and did not evolve separately. He proposes a
theory in which music serves a central role in early human mating, much as it does for some
species of birds. An individual who can spend effort on such a useless skill such as playing
music is surely pretty well off, demonstrating a high degree of intellectual and/or physical
prowess.
2. How can the knowledge obtained from this book be used to
create intelligent machines?

This books main focus is on musical perception, interpretation and, to a lesser extent, playing.
My view is that the quest to build intelligent machines is nowhere near the state of building
machines that should be concerned with any of these aspects, as there are far more other aspects
that have to be addressed first (just making a machine actually be able to hear for example).
Maybe in the far future, machines that can play music and musical instruments might get
built. In order to play music, they will have to be able to understand music, which is where some
of the ideas in this book could come in. To what degree these machines will be successful
compared to human players remains to be seen (see the argument above about conferring
emotion).
Another suggestion of where machines might take over is in any application that requires
song recognition. An example is the licensing agency where operators must monitor airplay of
songs. Algorithms for song recognition exist (the author himself worked in the field), but are
nowhere mature enough to fully replace humans, most probably because they dont consider too
many aspects of the song being analyzed.
Yet another application for musically trained machines is in personalized radio stations. I
have used a similar service like this in the past and it was pretty good (most of the times),
although Im guessing it worked based on previous categorizations of songs by human operators.
Machines that can perform this categorization automatically or even emulate a users preferences
would be a step forward.

Page 1


Essay
This Is Your Brain On Music
by Daniel Levitin

From Georg Victor
Introduction
The book This Is Your Brain On Music from Daniel Levitin is all about how music and your brain,
respectively your mind, are connected. The author is quite an interesting character, who started as a
musician, became a producer and started his academic career late in his life, compared to the majority
of scientists.
Since he never was an engineer and he never touched the topic of artificial intelligence, not during his
academic work and neither in the book this essay is about, there are much less connections to AI than,
for example, in the first book of the course, On Intelligence, by Jeff Hawkins.
Still, there are many ideas the two books share: Equality and consistency, for example. Also Daniel
Levitin states in different parts of the book that all sensory input is processed in a similar way. Also
talents or expertise in completely different fields is formed the same way: roughly 10,000 hours of
practice is necessary to become an expert no matter if we talk about scientists, athletes, artists, etc.
Main Point
On the other hand there is one fundamental discrepancy between the theories of Hawkins and Levitin:
the way our brain stores memories. Hawkins book is good for telling us how the neocortex is organized
and how we can build a system simulating the way we think and with that, intelligent machines. This
book, This Is Your Brain On Music, gives us a suggestion on how to organize the memory in an
intelligent way.
The author describes the two leading theories and how none of the two is completely right. In a very
convincing way he points out the strengths and weaknesses of each theory. The first one is called
constructivist. Like Hawkins describes it in his book, we humans only store the essential relations of
objects, experiences, etc. Since that, it is possible to e.g. recognize a song, even if it is distorted in one
or many ways. The second theory, called record keeping/tape recorder, lacks an explanation for this.
It says that all the details of our input are preserved and only lost due to normal biological processes.
But isnt this what we also experience each and every day, that we remember exact details of
something? Levitin calls many examples, e.g. the one experiment where people remembered the exact
pitch of their favorite songs. This experiment is particular good because Hawkins, who simplified many
things (deliberately to get a streamlined, easily understandable book that gives a consistent theory
about how our brain works the Memory Prediction Framework), stated that we just store the
relative change of pitch of a song, not the pitch itself.
Daniel Levitin develops an approach that combines the both theories in a new way. I would summarize
this approach as based on
Georg Victor Essay on This Is Your Brain On Music

Page 2


Hierarchical, prototype-based, context sensitive, dynamical categories
The author did not come up with this theory. Eleanor Rosch is a professor of psychology at the
University of California, Berkeley. She did her undergraduate thesis of her philosophy study about
Wittgenstein. Later on she quit with plain philosophy and finished her Ph.D. at Harvard with the
fundamental thesis about categories.
(A) Categories are formed around prototypes.
A category, e.g. fruit, contains several attributes, e.g. peaches, bananas, etc. But each
category contains one prototype. For fruit, this could be apple for most of us Europeans.
We divide the world into basic level categories that include as much as possible attributes.
The size is limited by the second principle of sharing as few as possible members with other
categories.
(B) These prototypes can have a biological or physiological foundation;
If we take the fruit example: Of course in other cultures, where there are no apples
available, there will be another prototype for this category. If the category even exists,
since categorization itself is of course culture, time, etc. dependent. Prototypes can also be
promoted by physiological facts, e.g. the anatomy of our ears.
(C) Category membership can be thought of as a question of degree, with some tokens
being better exemplars than others;
Better means here more privileged. How privileged a certain item is, can be estimated
by these three experiments:
Response Times Prototypes are elected faster in queries than non-prototype members
Priming Participants got triggered with a certain category. For prototypes, they
identified faster, if two words are the same (apple apple), than for other items
Exemplars When participants are asked to name items of a certain category,
better (more privileged / typical) items are named more often
(D) New items are judged in relation to the prototypes, forming gradients of category
membership; []
An interesting fact is that in some experiments of Rosch, she found out that we even make
up prototypes. In one experiment she showed distorted pixilated images to the
participants. These images were varying only a little bit from the prototype image. A
week later, they were shown again the images, now including these prototypes. A huge
percentage thought that they saw the prototypes before. So whenever we sense a new
item we ask ourselves (unconscious): how good does it relate to the (real or made-up)
prototype? for each category.
(E) There dont need to be any attributes which all category members have in common,
and boundaries dont have to be definite.
The borders of any category are not strict. Also, objects (attributes) can be the member of
more than one category.

Georg Victor Essay on This Is Your Brain On Music

Page 3


How can this idea help us building intelligent machines?
If we start asking the question from above, we only have to look at ourselves and our everyday live to
find the many advantages and surprising things we can achieve with the memory we have.
Of course there is the size of the memory. Nobody knows the exact number but Paul Reber, professor
of psychology at Northwestern University, states that we can keep information with an approximate
equivalent of 2.5 Petabyte. Sure, there are already supercomputers with this amount of storage
capacity and whats now huge will be found in personal computers in the near or middle future. Still, I
find this number stunning. It can only be achieved by the very organization of how we store
information. The brain uses the relation of objects, respectively sensory input - it does not store every
tiny aspect.
But much more than the sheer size of it is the organization and how it allows us to react in new
situations, how we interpret new sensory input. In the end it is this organization that really makes us
able to think the way we do. We can classify objects within seconds. We can distinguish easily if
different sensory input belongs to one entity or to different objects. Certain triggers bring back dozens
of memories, associations, and similarities with past events within fractions of a second. We can group
new experiences into the right cluster of familiar experiences and understand a lot through our
memory. We can even predict the future through our past at least partly. All of these points are until
now extremely hard to implement into machines and computers.
Personal Opinion
I personally quite liked reading the book. Unfortunately I started reading it too late (after New Years
Eve) so I had some time pressure and could not fully appreciate all the aspects of it. Still, it gave me a
good understanding of how complex the affects of music on our brain are and how deep it is involved in
our evolution.
It is funny to read that Darwin was equating music with the peacocks tail. But isnt that what we are
all thinking? That the guys playing the guitar or singing in a cool band (thus having the biggest tail)
always get all the girls? Now I finally have the scientific evidence. It has to be like that. No reason to get
intimidated or insecure. Fortunately the author points out many other peacock tail categories we can
choose from, if we are bad musicians: luxurious items, extensive dancing, general display of fitness, etc.
So there is hope for everyone!
It made me a little bit sad, or should I say frustrated, that Levitin states several times how our brain-
development is finished more or less with the beginning of our twenties. How difficult it is to learn
completely new things when we did not develop the brain structures before this time of your life. How
determined your preferences are. I mean, of course I know all of these, I heard it before. I dont believe
we can invent ourselves completely new just if we want it. But to read it over and over again black on
white is quite depressing, isnt it? Thank you, Daniel Levitin
Anyway, the thing that made me stop crying is what the book is all about: music. Levitin names so many
different bands and tells anecdotes about different artists that I started to listen to a lot of new music
while reading this book. I started to appreciate music and my Nubert loudspeakers, my Yamaha AX-497
even more than before. Thank you, Daniel Levitin

Anda mungkin juga menyukai