Press "Enter" to skip to content

Navigating our world with a mental GPS | Vijay Balasubramanian | TEDxPenn


[Music]
[Music]
[Music]
tell me what’s really good so hi I’m
Vijay Bala Subramanian and I’m a
theoretical physicist and I think about
this thing up there it’s your brain the
thing inside your head and I’m
interested in trying to understand the
rules and principles laws if you will
that govern how the circuits inside your
head let you live love learn and
understand TED talks
now um that’s a big topic and to
illustrate the kind of approach I take I
think it’s useful to have an example so
we’re going to talk about a small
example and the example we’re going to
discuss is how your brain maintains your
sense of place how you know where you
are in an environment now the brain at
its fundamentals is made of neurons and
neurons are cells like any other cell in
your body but they’re special cells at
the top in this picture behind me you
see these wires coming in the dendrites
those are the input wires of a neuron
that allow it to take in information
from other things like the environment
or other neurons will bring it into the
neuron the neuron then does something
figures out if it’s interesting and then
sends messages out through an output
wire which is the long wire coming down
from the neuron that’s the axon and that
sends messages elsewhere
now what sorts of messages are these
well neurons are electrical circuit
elements just like transistors and
diodes and what have you in silicon
circuits so neurons their body the cell
body is normally at a voltage of about
let’s say minus 70 millivolts
and then if the neuron find something
interesting in its input it goes boom
and makes a very rapid voltage spike
they’re called action potentials in the
trade and anyway you make this little
voltage spike and go Zook down the axon
and sends messages to other places so
all of your thoughts are really
electrical signals like there would be
in the computer but of course a single
neuron can’t appreciate poetry
can’t love somebody else and so all of
that must come from more and indeed
neurons are gathered together in two
circuits and networks of a very big size
in the brain so behind me there’s a very
famous and very beautiful picture of the
retina which is a piece of your brain
stuck at the back of your eye that
converts photons into particles of light
into electrical signals for processing
in the brain now this picture that
Ramona cahal the founder of Neuroscience
drew about a hundred years ago shows
many many kinds of neurons connected
together in some very complex Network
just as if you were to look at a
microchip you would find many kinds of
circuit elements connected together in a
complicated circuit so fine so there are
circuits of this kind but of course this
is a specific circuit it does one thing
it takes light and makes electrical
signals and if you think about the whole
variety of talks that you have there
must be more still and indeed there is
so this is the whole brain the whole
brain has different kinds of circuits
and networks embedded all over it that
do all the different things that we do
so for example right here on the back of
your head there’s the visual cortex that
has the circuits that allow you to see
and process visual images in the front
over here and your prefrontal cortex
there are the circuits that help form
your personality and help you with
things like planning and decisions and
things of this nature
today we’re interested in your sense of
place how you know where you are and the
circuits that help you know where you
are are stored in the middle of the
brain so in the middle can’t draw it so
easily in the picture there and in two
areas of the brain called the
hippocampus and the ento Reinold cortex
we’re mostly going to talk about the
internal cortex great
so that’s your introduction to
neuroscience now I’ve been talking about
your sense of place now what we mean by
that we should start by discussing what
you mean by your sense of space of
location so here’s a location a place
that I’m familiar with this is my living
room and I know it very well when I’m in
it I know I’m there I can close my eyes
and I can walk about and basically not
bump into stuff
somehow somewhere inside my head here
that place is stored and we’d like to
know how now how is here stored well I
already told you that everything in the
brain is made of neurons firing
electrical signals so somehow this place
must be an abstract pattern of neural
firing inside your head and somehow that
abstract pattern of neural firing must
maintain a map of your location in the
trade that map is called the cognitive
map and people have been wanting for a
long time to understand this now about
12 years ago a very important clue
developed in our understanding of this
cognitive map so there was a land in
Norway of Edouard Maybrick Moser and
they had rats running around some
enclosure and they had implanted
electrodes to record from a piece of the
brain the n-terminal cortex which has
been misspelled there
the enter rival cortex and um the arm
and what did they find
well they would record from a neuron and
they would find that the neuron would
fire when the rat was in specific
locations in this room so Gobert stop
murder stop as the rat walked run so
they kept that they collated all this
firing and up top in the topmost picture
you see a hotspots of where this neuron
fired or where the rat was with his
neuron fired and what do you see you see
that this particular neuron fired when
the rat was in this enigmatic grid of
locations in the air inside the room
even more surprisingly or interestingly
as they poked deeper into the internal
cortex with their electrode they found
other neurons that fired on bigger grids
and then on bigger grid still so there
were these lattices in the world arm
which at which these neurons fire so
here’s the question Oh what’s more even
more interestingly by the way if you
took the rat away right and brought it
back the next day this kind of map if
you like of the world he’s kind of grids
embedded on the world were uploaded were
reloaded in to the rat’s brain
instantaneously so it really felt like
this is some kind of map of the world
but the question is how right what do
these patterns mean but how did they
maintain a map of the world now I’m a
theoretical physicist and my job in life
is to dream up explanations for
interesting things that happen in the
world that’s what I like to do for a
living
that’s what I do for a living now that’s
a trade that’s a job that requires two
important traits one is imagination
because you need to dream up new
explanations for things and it needs
courage because you can often be wrong
okay but nevertheless let’s try let’s
try to imagine an interpretation of
these experimental findings okay so I’m
going to follow the method the
physicists usually follow which is to
first simplify the problem right I’m
going to imagine that we live in a
one-dimensional world that’s the line
you know we live on a line and that the
whole world is eight meters long and
that for whatever reason we need to know
where we are with a resolution of
precision of one meter so I want to
imagine how I would devise a neural
system to tell me where I am with that
position
well easy right I’ll break up the world
into eight blocks like so up there and
then I’ll assign one neuron the fire in
each of those eight blocks so as you
move around if your neuron number three
you fire when they’re Rath isn’t block
number three that kind of thing well
clearly that’s a map of the world right
it’s a kind of representation of where
you are and it requires eight neurons so
here’s another way what you could do is
you could take the whole room and divide
it into two halves I’ve colored that’s
the top line over there and I’ve colored
the left half blue and the right half
orange and we get the blue neuron to
fire when you’re on the left half of the
room and the orange neuron to fire
when you’re in the right half of the
room then you divide the room into
quarters okay and now I again assigned
two neurons want to fire along the
second line there want to fire in the
left half of the left half of the room
at the left half of the right half of
the room and the aura and another neuron
the orange one to fire in the right half
of the left half of the room at the
right half of the right half you get the
picture
you subdivide by twos okay now suppose I
was going to call these neurons the zero
neuron under one neuron based upon left
versus right then the first two location
the one on this side would be labeled
zero one zero and the second location
marked the second dashed line would be
labeled one zero one clearly this is
also a way of maintaining a map of the
world if you were a computer scientist
you’d probably do it this way and this
would be what’s called a binary code for
the world in a binary because of zeros
and ones and using binary codes to
number out all the locations if you
wanted you can use divide into tens and
make a decimal code okay so these are
the ways we can imagine writing out your
location in one dimension so let’s apply
this experience to thinking about how
you would map out your location in two
dimensions because that’s what we seem
to do that’s what the rats seem to do as
they walk around the world so here we go
imagine again in a simplified way but
the world is an eight meter by eight
meter box and that I need to know where
I am with a precision a resolution of
one meter by one meter that’s what I
need for my behavior how could I do this
but following what we imagined before I
could break up the whole world into 64
little boxes that are one meter by one
meter
assign one neuron the fire in each of
these little boxes and I would have a
map of the world it would cost you 64
neurons but you could definitely do this
so that’s one thing we could imagine
here’s another thing we could imagine
I’ll start but breaking up the world
into four corners Northwest northeast
Southwest southeast all colored
differently here
then I’ll take let’s say the northwest
corner
and break it up into quarters then take
each of the species and break it up the
quarters this is just like what I did
before if it was a one-dimensional world
right and then if I wanted to tell you
where the cross is in that picture I
would say it’s in the blue yellow blue
right that’s a way of telling you where
you are so that is one way of producing
another kind of map of the world and of
course you could subdivide the tens like
decimal systems on all these kinds of
things so this is a way of describing
where you are so I would now like to
draw an analogy look at these pictures
on the left side we have the
experimental data on the right side is
the thing we just imagined as a way of
describing position don’t they look
similar first at one end you have big
grids then you have smaller grids then
you have smallest grids and in some way
you can put this information potentially
together so if you believe the analogy
we would say that somehow the brain the
internal cortex has developed a sort of
2-dimensional fuzzy sort of number
system for numbering out locations in
the world that would be the
interpretation you would place on this
so that’s a curious talking how can you
build evidence for this right to build
evidence for a theory you need to make a
prediction that you can test so let’s go
for that so how do you build such a
prediction well if this is like a number
system I could ask should the brain
build a binary number system should it
build a decimal number system what
should it do and in this neural context
whether it’s binary or decimal really
what that’s really talking about is you
see how these grids get bigger and
bigger and bigger well you can take the
ratio of the size of one grid divided by
the size of the next smaller grid
telling you how to scales up and that’s
really telling you what kind of number
system you would have so if it scales up
by factors of 2 it’s like binary if it’s
like fales up plus scales up with
factors of 10 then it’s like decimal so
we could try to predict this but if you
really believe our theory of the brain
that we’re building here we should be
able to predict
what sort of number system the brain
should use to represent location in a
2-dimensional world for that Li the
principle what principle can you use
well so here’s the idea so I’m saying
the brain should have a certain kind of
two-dimensional number system to
describe the world well here’s a
principle that neural circuits should be
organized to consume resources
efficiently while I saying this well the
brain you know on the one hand is
extremely efficient it consumes about
twelve to twenty watts of power whereas
your laptop consumes about 80 watts of
power and you know your laptop doesn’t
write poetry
right so it’s somehow extremely
efficient on the other hand your brain
is also extremely expensive it’s 2% of
your body weight but fully 20% of a
metabolic load your brain costs you more
than muscle when you’re working out so
this is lead neuroscientists to suppose
the neural circuits should be efficient
in their consumption of resources
because they’re already so expensive
that there should be something in their
organization that makes them efficient
in using up you know power and space and
so on in the head so we can hypothesize
that the grid system that I told you
about in the head is in number system
and that it should be organized to
minimize the cost of representing where
you are in the world okay so physicists
purists like me what do we do when we
see a thing like this
we mathematize it you know make it more
abstract so here’s an abstract
description of the grid we put down some
variables into parameters and we build a
theory we say okay the resolution the
precision of this abstract description
of the world should be set by the
behavior of the animal we say that the
cost of the grid of the system should
grow with the number of neurons because
surely the more the neurons you need the
more expensive the whole thing is going
to be and that will agree that what we
want to predict is how this thing scales
up the ratio of the sizes between these
scales okay and then a good theorist can
write down a mathematical theory of this
and ask which grid which way of
representing space of location of the
world is the most efficient
so you can do that and what we found is
that a prediction the prediction is that
the way in which these grids will scale
up as you poke into the internal cortex
is by ratios of the square root of a
famous mathematical number called e this
e is Euler’s number a number that Euler
the famous mathematician devised for
other purposes and what do you know
experiments confirm this theory
precisely so what do we learn from this
well I think at the simplest level we
learn that nature seems tough long
before humans did devise these kinds of
complicated number systems as and in a
two dimensional fuzzy neural kind of way
and then proceeded to optimize it for
the neural hardware I think that’s the
first lesson we learn somewhat more
broadly these results suggest the brain
is a sort of elegant and efficient
mathematical machine and that we can use
this idea of elegance and efficiency of
this engine in our head to build a
theory of organization of the brain even
more broadly I think what I just told
you has a bearing on a famous dictum of
Galileo so hundreds of years ago Galileo
wrote that the book of nature is written
in the language of mathematics and you
know physicists people like me have
especially taken that seriously so over
the decades over centuries there’s a
fully mathematical description of the
world and physicists believe you have
explained a phenomenon if you have a
precise mathematical description of it
and that makes predictions for the next
experiment I believe that this is the
century where the living world will
receive finally the same kind of
treatment for you know again for
centuries we’ve described life and
living things mostly descriptively but I
believe that in this century the study
of life will become as mathematical and
beautiful as a study of atoms crystals
and the forces of nature thank
I
Please follow and like us: