Understanding your brain as a network and as art | Danielle Bassett | TEDxPenn

Translator: Sara, English-Persian (Farsi)
Reviewer: Denise RQ I really, really like complicated things. I love complicated flavors, like putting avocados
on coffee ice cream, It’s a little bit bizarre,
but actually kind of good. I also really like complicated music, so give me Mahler over Mozart any day. I love complicated plots. Sit me down with a good
Thomas Pynchon novel. In fact, nerd alert, I actually have
a computer named Pynchon. But, we are all actually part
of multiple complicated things. And the complicated thing that I like to think about
the most is networks. So we are all part
of multiple, ever changing networks. For example a network
of colliding particles in the Universe, a network of interacting flows
in the Earth’s atmosphere, a network of highways and city streets
on the Earth’s surface, and a network of family, friends,
coworkers, and neighbors. But probably the most complicated network
that you engage with in your everyday life is the network inside
of your head: your brain what makes you who you are. So how your brain is a network,
and what is a network to begin with? Well, a network is a way of looking at a system composed
of two different kinds of parts. The nodes of a network
are the sub-parts of the system, and then the edges of a network are
the interactions between those sub-parts. So what we do as network scientists is that we take a system
that’s pretty complicated, and we separate it out into these nodes, the little bits,
and the edges, the interactions, put them together, and try to understand how that combination creates a structure
that enables a system to function. So we take these representations,
and we try to understand where are the parts of the system that are connected up
to a lot of other parts. Those are hubs of the system, and are often really important
in the functioning of the system. We also ask, where are there
a bunch of nodes that are densely interconnected
with one another. Those are called modules or communities, and they can often perform
a single function themselves. We can also identify lonely nodes
that are not very well connected. They’re actually called
peripheral regions, and they tend to have
very peripheral functions inside of the network as well. So, if this is what a network is,
then how is your brain a network? Well, first we have to think about
what are the network nodes? If I say your brain is a network
you might be a little confused because the brain is actually
a thin piece of tissue, right? It’s an organ, it’s a volume. It’s not composed of little pieces,
at least by eye. But actually, if you look
at the surface of the brain, it’s composed of these peaks and valleys. And the peaks and valleys
are called sulci and gyri, and they help to define
particular parts of the brain that have different kinds of neurons
and different distributions of neurons, and therefore, can have
very specific functions. So what we can actually do is we can use
the surface morphology of the brain to identify independent pieces
that may perform different functions. Then the question is,
if those are the nodes in our network how do those pieces interact? What are the connections that underlie the passage
of information between those areas? To answer that question what we have done is used a new technique
called diffusion imaging. So, diffusion imaging is something
that you can take with an MRI machine, and it actually watches water molecules
bounce around in your brain. You might have not known water molecules are constantly bouncing
around in your brain, but they actually are. And what they do is that they bounce against walls,
they bounce up against these walls. The walls are actually
big bundles of neuronal axons. So by watching
where the water molecules bounce, you can infer where the bundles
of neuronal axons are. Those neuronal axons
are actually the highways along which information
can propagate in your brain. So these are, in fact,
edges to some degree. These are interactions,
these are pathways. So when we initially saw this kind of data
it looked really messy and really tangled, and we were wondering, “How could we actually understand
this structure, and how can we compare it
between different individuals?” Network science actually provides us
with a way of quantifying that in a really beautiful summary statistic. So we were able to quantify
individual nodes that have specific functions, and then map their interactions
using diffusion imaging. So then the question is, is network science something
that is actually intuitive to us and can we learn something new
about the brain using it? The answer is yes. So when I talk about networks,
it might actually bring you back to grade school
or maybe preschool, actually, where you played with toys
that linked up together. Here’s an image of tinker toys. So, tinker toys are created
from disks and rods, and it’s actually the combination
of these rods and these disks that make a network. So if you watch a little kid play
with one of these, what happens is that they put
the little disks and the rods together, and they build up a beautiful structure
– a sculpture, to them – that is a network sculpture. But then they don’t stop there, right? They actually usually take a piece off
and maybe put it somewhere else, or maybe they completely disassemble
the structure, and then reconfigure it. And then maybe they disassemble it again,
and reconfigure it again, right? So actually what they’re making is many different networks
out of the same underlying components. And, in fact, that’s exactly
what your brain does as well. So your brain takes the pathways
that exist in there, and uses different pieces of it
in different network configurations to enable you to perform the functions
that you perform in everyday life. So every task that you perform,
every thought that you think, every thing that you do, requires a different set of patterns,
a different set of connections, and that’s how your brain
actually distinguishes between the different tasks
that you have to perform. What’s interesting about this is that this is actually different
between each of us, each of us has a different kind
of reconfiguration pattern. And how did we discover that? Well, we discovered that
in the context of learning. Learning is a scenario where we have
to actually change our brains in order to change our behavior. So traditionally, if I was to say
that I wanted to understand learning, what I would probably do is take
two portraits, two still photographs. I’d take a picture of your brain
before you learned, and then I’d let you learn, and then I’d take a picture
of your brain after you learned. And then I’d compare the two pictures
and say, “Are they different?” If they are different, great. That mean that your brain actually
did change and you were able to learn. But that never really felt
very fulfilling to me. What I really wanted to know was not just whether or not
the brain changed, I wanted to know how it changed. So instead of two still photographs, I want to be able to have
a series of snapshots that follow the brain as it reconfigures. So can we do that? The answer is yes. What we do is that we use
MRI scanning techniques. We put healthy individuals
inside of the MRI scanner, and have them learn new skills. The skill that we love to use
is actually a set of finger sequences. So, they see a pseudo-musical staff,
similar to a guitar hero, and then it tells them
which buttons to press, and they press that series of buttons,
similar to playing piano arpeggios. And as they’re doing this,
we’re imaging their brain and watching exactly which patterns occur,
and how they transition over time. So, what this gives us
instead of two still photographs, is a series of snapshots
that quantify that reconfiguration. But again, it really wasn’t
quite what I wanted. Because when you look at this kind of data
it looks a little bit like a documentary or, a slide show,
similar to the one I’m giving now. What I really wanted was to have a video, a way of seeing the smooth transition
in the brain as it reconfigured. Can we do that? Well, yes. We use a mathematical trick, that is, actually, to link up the networks
in individual time slices. We take two minutes of your learning
and link it up to the next two minutes, and the next two minutes,
and the next two minutes. In network speak, this is actually
using a linked adjacency tensor, which quantifies, mathematically, the object that is a smooth video
of your reconfiguration over time. So, this critical change
in an underlying object enabled us to see how people were adapting
and changing their behavior. What did we learn? Well, we learned that there are two parts
of the brain that are critically important in these kinds of motor skills. Number one is the motor cortex,
highlighted in red, which is what’s necessary
for actually moving your fingers. Secondly, the visual cortex,
which is highlighted in yellow, is important for actually processing
the guitar hero pseudo-musical staff. And what happens is
that initially during learning, these to parts of brain
are densely interconnected. Why? Well, because they have to
share information with one another, and they have to communicate
with one another. They have to take in
the visual information, and then process it
into a response in our fingers. So they have to communicate. But as time goes on, and as people learn,
we found a surprising fact: that these two pieces of the brain
stop communicating almost altogether. They become almost autonomous
from one another. So, why is that? Well, think about when you played
an instrument or when you learned a sport, or any type of motor skill
that you have learned in your life. Even walking, for example. That’s something that you have learned
so well that you no longer think about it. It’s put down into motor memory, and you can just do it automatically
without thinking. That’s exactly what happens
in learning these finger sequences. And we can actually see that
in your brain: the two pieces of brain
stop communicating with one another and are able to just
respond automatically. But then my question was, if this is true, and this reconfiguration process
is happening continually as we learn throughout our everyday lives, could I actually determine who was good at learning
and who was not so good at learning? Could I predict from your brain data
whether or not you’d be a good learner? And the answer is yes. And this was really surprising to us. So the people who are actually
good learners are the people whose brains reconfigure really swiftly. And the people who are not
particularly good learners, are people who have really rigid brains, and the functional connections
between different brain areas really don’t change very much
during the time period of the experiments. This fascinated me, because I started
asking the question about myself: is there a way for me to change
the flexibility in my brain so that I can learn better? Maybe if I had
a little bit of coffee this morning, or maybe if I slept better last night,
would I be able to be more flexible? That’s an open question,
and the one I care about, but I think the implications
for this finding is actually more broad. So, let’s imagine rehabilitation
after a stroke, for example. Would there be a way for us to determine who and when to train based on
their brain network flexibility? Or let’s think about interventions, therapeutic interventions
or stimulation interventions. Is there a way for us to determine
where in the brain we should stimulate to enhance flexibility and therefore,
enable enable learning and rehabilitation? Then, let’s get a little crazy
and think ten years down the road. In a classroom environment,
would I be able to determine which classroom environment is ideal
based on the amount of flexibility that classroom environment engenders
in little developing brains in children? So, would I be able to say
this classroom environment is really excellent
because it engenders flexibility in brains and this classroom environment
is not as strong because it does not engender
flexibility in people’s brains? These are areas we’re extremely excited
to follow up on in the future. But I actually have
a more pressing problem today. And that is, understanding
and quantifying the structure in these reconfiguring networks. Most of the illustration that I’ve shown
you today are just that, illustrations of the data that we see;
they aren’t the data exactly. And the reason that is, is because
this data is really complicated. It’s very hard to understand the structure
inside of it; it really is big data. Traditionally, if I wanted
to understand this, I might turn to computer science
or machine learning to extract structure from that data. But we’ve actually decided
to take an alternative route, and that is to speak with people who are trained to identify structure
in messy data, and those are artists. So, we are actually formalizing
collaboration with professional, undergraduate, and other local artists
in our community, to help us to understand the networks that we are studying,
and the data that we’re trying to analyze. Here are a few examples of some of the pieces
that have been created over the past year. These are examples
that quantify network organization that is being addressed and trying to be
understood over the entire Penn campus. And what we’ve seen
by this collaboration with the artists is that we begin to identify
new areas of research development. So here’s a piece
that I really wanted to show you. And that is a piece by a fashion designer,
Katherine Khorassani. What she did is that she quantified
the reconfiguration of the brain networks during epileptic seizures. So what she did is she used
Verizon telephone wire to illustrate the communication patterns
in the brain, and then she knitted them so that you could see
the undulating pattern of brain rhythms. Following that, she sculpted
this circular architecture to illustrate the evolution
of the network seizure. What we were really inspired by,
is to quantify the volume, or the energy, or the burstiness of these networks,
in new ways, mathematically speaking. So this artistic representation directly
informed new hypotheses in our group. But we don’t actually stop there. We take these pieces
into local classrooms, to encourage children into interaction
between the arts and the sciences, and to illustrate the concepts
of networks in their everyday lives. And even in this scenario, we’ve actually been able to take away
new ideas for our research. I like to highlight this one, where an eighth grade student
actually shows the structure in his brain network
on the black cardboard. And then behind that, he shows
many, many colorful squiggly lines, which illustrate the chaos or the noise
that he feels in his brain. So, one of my questions is,
does noise actually impact on flexibility? Does it enhance flexibility
or decrement flexibility? And how can we empirically test
that relationship? Looking to what lies ahead;
I think that what lies ahead is the ability to tune our brain networks
to optimize our cognitive processes. And for me as a scientist, I’m curious,
as I collaborate with artists, is my brain becoming more flexible? well, I don’t know the answer, but it sure
is an exciting adventure in the meantime. Thank you. (Applause)

8 thoughts on “Understanding your brain as a network and as art | Danielle Bassett | TEDxPenn

  1. good work i posted you on facebook…. lol…. im tying the quantom platforms as we speak…

    much love baby girl.

    god bless you…

  2. This seriously got my heart pumping! What you said about trying to make your brain more flexible resonates with me! I am going to have to look more into this

  3. Could you link up the brain to a computer by connecting just a few of my human neurons to some external machine network ? , What if you just put in an empty network and let your own brain fill it with data ? Do certain shapes of networks just make you smarter in the human brain ? Why are some people really smart, is it the network structure rather than the data that allows higher intelligence? Is there a network shape in the brain for improved creativity or improved motor skills or better memory ?

Leave a Reply

Your email address will not be published. Required fields are marked *