[CS] Bharucha Lecture Notes

[Elton Joe]

Some of my notes were crappily worded:

Yes, I meant "musically untrained" when I wrote "uneducated" (couldn't
think of the word).  Although, I wonder what this really means.  There
are so many different ways of learning music, and I know that some
people who don't read Western notation can play amazingly, while some
people who do read fluently can't "think outside the box" and learn to
improvise.

And when I mentioned singing, I did not mean the subjects actually
singing in the MRI machine themselves, I was just mentioning the fact
that all of the music samples had singing in them and wondering if
singing was speech or music or both or what?


elton

On 2/24/06, Jaime Davila <jdavila at hampshire.edu> wrote:
>
>
> Elton Joe wrote:
> > Here are my notes from the Bharucha lecture, in case anyone wants
> > them.  It's paraphrases mixed in with my own questions and comments.
> >
> > My main question is one of basic neuroscience:  What happens,
> > biologically, when part of the brain "lights up," and what is going on
> > (or not going on, in comparison) in the rest of the brain?
>
> It's measuring blood flow. The more blood that is flowing into an area,
> the more that area lights up. All of the brain areas have blood flow,
> but the flow increases in some of them.
>
> >
> >
> > elton
> >
> > MUSIC, BRAIN, AND CULTURE
> > Jamshed Bharucha
> >
> > Former editor, journal "Music Perception"
> >
> > It has often been said that culture and brain biology are two separate domains.
> >
> > However, we "carry our culture in our brains".
> >
> > We have only begun to talk about culture through cognitive science,
> > but it will be the norm in 10 years.
> >
> > Slide of MRI brain scans with "activity" marked.
> > "Activity" is "greater than chance".
> > So what is activity, and what is chance?  What is going on in the
> > areas of the brain not marked, and why is "less" going on there?
> >
> > Neuroscientists strive for more accuracy in measurement (voxel-size
> > and temporally).  What would greater accuracy show?
> >
> Right now, when you see one of those color spot in an MRI image, the
> increases blood flow might be in one piece of that box but not in
> others. Or it might be in the whole box. The system doesn't have enough
> precision to distinguish. More accurate machinery would allow for a more
> detailed image. Would that help? Maybe. It sure wouldn't hurt.
>
> > Auditory cortex is marked by greater activity during the listening of
> > music.  (Machine noise is filtered out statistically - it does
> > activate the auditory cortex as well.)
> >
> > These particular scans show activity localized to the left hemisphere
> > side of the auditory cortex.  The norm is on both sides.
> >
> > Study:
> >
> > Participants listened to a particular melody in all 24 major and minor keys.
> >
> > The music you normally hear within a culture is highly constrained in
> > terms of transitions from one key to the next.
> >
> > Transitions - C maj -> G maj is more frequent than C maj -> F# maj
> >
> > Found that uneducated listeners were still able to distinguish key transitions.
> >
>
> You mean musically untrained, right?
>
> > Which part of the brain is correlated with usual/unusual key change?
> >
> > Hypothesized that the auditory cortex would be involved, since it
> > lights up during music listening.
> >
> > Frontal cortex lit up instead, which indicates that there may be
> > cultural factors involved.
> >
> > Idea:  brain internalizes culture automatically by being embedded in it.
> >
> > Study:
> >
> > Group of college students with little experience with Indian culture.
> > Group of college students who grew up in India and had exposure to
> > Hindi and Indian music.  (bilingual and bicultural musically)
> >
> > Western and Indian classical music
> > Western and Indian pop music
> > English and Hindi speech
> >
> > Participants listened while in an MRI machine, but were not asked to
> > attend to a particular task (this would be used in an ideal
> > experiment).
> >
> > What about singing during the music?  Wouldn't this activate music and
> > speech at the same time (assuming the brain processes them
> > differently)?  Is singing music or speech?
>
> You start singing, a bunch of other stuff has to come into action.
> Including those areas of the body that deal with facial movements.
>
> In terms of experiments that look into this, I think that would be
> enough movement to keep the MRI machine from getting a decent image.
>
> >
> > Brain vector - a set of numbers that indicate the level of activity in
> > each of the areas measured.
> >
> > Brain vectors were paired up to calculate the similarities between participants.
> >
> > Multi-dimensional scaling (statistical analysis) was used to place
> > participants in an abstract mathematical space.
> >
> > Brains which were similar were close together, and brains which were
> > dissimilar were far apart.
> >
> > "Cultural space maps on to brain space"
> >
> > Classical music (Western/Indian) shows distinct separation between the
> > two groups
> >
> > Used vocal music to keep the voice aspect constant, but Western
> > classical music was in non-English languages (Italian, German, etc.)
> > and Indian classical music was in non-Hindi languages (such as
> > Sanskrit).
> >
> > How was classical music defined?  Similar to the problem of defining
> > speciation, how does one distinguish between variation within a
> > particular style vs. variation between styles?
> >
> > NEURAL NETWORKS
> > How might the brain predict pitches according to cultural bias?
> > (simulation project)
> >
> > Layers of neurons with connections between layers.  Connections have
> > particular strengths.
> >
> > If a particular sine wave is played, particular neurons light up.  In
> > other words, particular parts of the auditory cortex are tuned to
> > particular frequencies.  (First layer neurons are innately tuned)
> >
> > Second layers correspond to some sort of cultural conditioning.  The
> > neuron in the second layer that lights up the most due to a set of
> > stimuli from the first layer is the "winner" - it corresponds to this
> > particular stimulus the most and gets stronger.  (Uh oh, grandfather
> > cell - what happens when you need to represent different things with
> > the same neurons?  Can one neuron do more than one thing?  What
> > "measures" the whole process?)
> >
> > Hebian learning - essentially the neural network connection strength theory
> >
> > Strengths of the connections (synapses) can be plotted on
> > multidimensional graphs.
> >
> > Theory applied to cultural bias building:
> > When a new stimulus is presented, the vector that it is closest to is
> > reinforced or tuned to the cultural environment.
> >
> > "Self-organizing models" - Bharucha says this is how the brain MUST
> > and WILL learn cultural bias.
> >
> > Indian scales - they sound foreign to the Western ear.
> > There are scales (such as Lydian) that fell out of usage by Western music.
> > (although they have been picked up by jazz)
> >
> > FLATTED SECOND
> > The flatted second is common in Indian scales, but not in Western
> > scales.  Therefore, the Indian brain would infer that the flatted
> > second was present if all of the other scale tones were played.  The
> > Greek modes have some flatted seconds.
> >
> > The brain fills in musically from fragments in order to make a
> > complete scale it can recognize.
> >
> > You can move through your environment more easily if you can make
> > predictions about what to expect culturally.  This is similar to
> > prototype theories of aesthetics, and in general the idea of schemata.
> >
> > MOZART EFFECT
> > Theory was that Mozart contains certain intervals that make you smart.
> >
> > However, when only one voice was changed by half a step, there was
> > absolutely no effect.
>
> I think you might not be stating that correctly. What he said was there
> was no difference between people listening to the original Mozart and
> the distorted version. Had the original causes improvements but the
> distorted version failed to cause improvement, it would be corroborating
> the Mozart effect.
>
> >
> > ORIGIN OF MUSIC
> > pre-linguistic auditory communication
> > music is closely linked to movement (dance, etc.)
> >
> > TALENT
> > Is talent due to prior tuning of the neural networks?  How does it happen?
> >
> > _______________________________________________
> > CS mailing list
> > CS at lists.hampshire.edu
> > http://lists.hampshire.edu/mailman/listinfo/cs
>
> --
>
> ******************************************************
> Jaime J. Davila
> Assistant Professor of Computer Science
> Hampshire College
> School of Cognitive Science
> jdavila at hampshire dot edu
> http://helios.hampshire.edu/jdavila
> *******************************************************
> _______________________________________________
> CS mailing list
> CS at lists.hampshire.edu
> http://lists.hampshire.edu/mailman/listinfo/cs
>