NANCY KANWISHER: So seeing
where animals are going, so you can avoid them if
they're coming after you or so you can catch them if
you're going after them, right? One of the arguably
uniquely human abilities is precision throwing, right? No other animal can do that. That's a very human thing. Although, visual motion is
shared with lots of ability to see motion is shared
with lots of animals. What else did you notice? What else seemed funny or harder
to discern with stop motion? Yeah? AUDIENCE: We care about small
details like [INAUDIBLE] to understand what
the person is seeing. NANCY KANWISHER: Yeah. Yeah, so I was
making notes to self. I haven't done that demo before. But in future, it would
be really good to have the audio quality terrible. Because if the audio
quality is terrible, you would lean more
on lip reading. And we might have noticed more. But it's really hard to
do that probably even at relatively fast flicker rates
because that motion information is important. Absolutely. What else? How about beyond
just lip reading? What else did you notice about
the faces, mine or Jim's? Could you-- yeah? AUDIENCE: They were static. So it was kind of hard
to tell like emotion because a lot of the
ways we express emotion is very nuanced. NANCY KANWISHER: Exactly. Exactly. Facial expressions
are incredibly subtle. Like little
microexpressions flicker across the face in a tenth
of a second and go away, and you guys detect them. Like we're very, very
sensitive to those things. Sometimes if you see somebody
in a hallway and, for a moment, there's an expression that
flickers across their face and then they give
you a normal smile, but you can tell
from that expression that actually they
didn't want to see you, for whatever reason, right? We catch those things. We're really, really
good at catching those little
fleeting expressions. And those probably have to
do with not just sampling with fine temporal frequency
but probably seeing the direction of motion of
each little part of the face. OK? OK, so this is just
common sense reasoning about what we might
have motion for. OK? And so you guys got all the
things that I had in mind. OK, so now the next question,
just kind of thought question, speculation question, given
these many different things that make motion important
to us, biologically, ecologically, in
our daily lives, maybe that's important
enough that we might allocate special brain
machinery to processing motion. What do you think? Important enough? Could you get by if you lived
in a strobe world all the time? Could you survive just fine? Hard to say, right? Might be hard. I mean, we probably don't need
to go hunting down predators. But you walk across
Vassar Street. And there's some pretty
dangerous predators coming down Vassar Street
in the way of cars, right? You need to know
where they're going and whether you can
cross in front of them. So it's actually pretty
hard to live life without being able
to see motion. And I'll tell you about a woman
who has that experience later in the lecture. OK, next question,
just think about this. I'm not going to test
you on it or anything. It's not the topic
of this course. But it's a perspective
you should take. Imagine that this
were a CS course and I gave you a
segment of video. And your task was to
write some code that takes that video input
and says whether objects are moving in that movie or
says which objects are moving or how much they're moving or
what direction they're moving. What kind of code
would you have to write to take that video input
to try to figure that out? OK, so just think about that. We're not going to be
writing code in this class. But a lot of what
we're going to be doing is thinking about,
how do you take this kind of perceptual
input and come out with that kind of
perceptual inference? And what kinds of
computations would have to go on in between whether
those computations are going on in code that you guys write
or in a piece of brain that's doing that computation? And thinking about how
you might write the code gives you really
important insights about what the brain
might be doing. OK? All right, so that's the
point of all of that. The Marr reading talks
about all of this. And the key point we're
trying to get here is that you can't
understand perception without thinking about what
each perceptual inference is necessary for ecologically
in daily lives and about the computational
challenges involved in making that inference. OK? So we'll get back to all
that next week and beyond. But meanwhile, here's
the agenda for today. So here's the agenda. We just did the demo. We're now going to skip and do
some neuroanatomy, absolutely bare basics. Because on Wednesday, we
have this amazing opportunity to have one of the most famous
neuroscientists in the world do a dissection of
a real human brain right here right
in front of you. It's going to be awesome. And I don't want to
waste that opportunity or embarrass ourselves
by having people not know the bare basics. So we're going to
do the bare basics. It's all stuff you should
know from 900 and 901. And I'm going to
whip through it fast, so we can get to more
interesting stuff and get back to visual motion. OK? That's the agenda. All right, so some
absolute bare basics of the brain, the human brain
contains about 100 billion, 10 to the 11th neurons. And that's a very big number. That's such a big number it's
approximately Jeff Bezos' worth. Well, it was until Mackenzie
got into the picture. So we'll see. No, you don't need to
remember this number. Just know it's a
really big number. Basics of a neuron,
here's a neuron. A neuron is a cell like
any other cell in the body. It's got a cell body
and a nucleus, just like any other
cell in your body. But the thing that's
distinctive about a neuron is it has a big long
process called an axon. It's got a bunch of dendrites,
the little processes, the little thingies
near the cell body. And out at the tip of the axon,
that's your classic neuron. Many neurons have
a myelin sheath, a layer of rolled up fat
around the axon made up of other cells. That makes the axon conduct
neural signals faster. OK, you should know all that. I'm not trying to insult
your intelligence. I'm just trying to make sure
everybody's with the program here. OK, so you have thousands
of synapses on each neuron. And that means you have-- to
put it technically-- a shitload of synapses in your brain. OK? Another important point, the
brain runs on a mere 20 watts. And if you're not impressed
with that, reflect on the fact that IBM's Watson
runs on 20,000 watts. So one of the cool things
about the human brain is not just all
the awesome stuff that we can do that
still no computer can do, that I talked about
last time, but also how incredibly
energetically efficiently we do it with our human brains. So most of this course is
going to talk about the cortex. That's all the stuff on
the outside of the brain. That's that sheet
wrapping around the outside of the brain,
that folded outer surface. It's approximately the size
and area of a large pizza. But there are lots of
other important bits too. And I'm going to just
do whirlwind tour of those other bits now. OK, so you can
think of the brain as composed of four major
kinds of components. Deep down in the
bottom of the brain, you have the brain stem, where
the spinal cord comes in here. And the rest of the
brain is up there. And the brain stem
is right down here. And the cerebellum, this
little cauliflower like thing that sits out right back there. And in the middle
of the brain, you have the limbic system
with a whole bunch of subcortical regions. And we'll talk about a
few of those in a moment. And you have white matter,
all the cables and connections that go from one part of
the brain to another part. This is an actual
dissected human brain. And all those kind of
weird fibrous things are bundles of axons connecting
remote parts of the brain to each other. You can see them in
gross dissection. OK? And of course, you
have the cortex. OK, so these are just four
major things to think about. And before we spend the
rest of the course on that, we're going to do just
a teeny little bit on the other major bits. OK, and I'm going fast. So just stop me if any
of this isn't clear. All right, so the reason
we're doing this in part is that, with a
dissection of a brain, some of the main
things you see are those subcortical
structures, right? And so even though the course
is going to focus on the cortex, each little different bit of
the cortex to the naked eye looks like any other
bit of the cortex. It's the subcortical stuff
that looks different, right? So that's why we're doing this. OK, bare basics
on the brain stem, you can think of it
as a bunch of relays in here, different centers that
connect information coming up from the spinal cord and send
it through into the cerebellum. So it's, in many ways, the most
primitive part of the brain. That means it's
shared with animals that branched off
from us very far back in mammalian evolution. But it's also essential to life. OK? So you can get by with
most of your cortex gone. Like you may not
have a lot of fun. You may not really
know what's going on. But you will stay alive. But you can't get by without
your brain stem, right? It controls all kinds of basic
crucial bodily functions, like breathing, consciousness,
temperature regulation, et cetera. So it's not interesting
cognitively. But it's crucial for life. Cerebellum, this
beautiful thing here, it's basically involved
in motor coordination. But from there on out,
there's a huge debate about its possible
role in cognition. And so there's lots of
brain-imaging studies where people find that
the cerebellum is engaged in all kinds of
things from aspects of perception up through aspects
of language understanding. You can find activations
in brain-imaging studies. Nonetheless, the best
guess is that you actually don't need a cerebellum
for any of this. So if anybody's
interested, I'm going to actually try to
remember to put it up as an optional
reading on the site. There's a recent
article in The Atlantic or The New Yorker about a
kid who had no cerebellum. And he learned to
walk late and slow. Nobody knew what
his problem was. But he learned to do
pretty much everything. Like he's pretty much fine. His motor coordination
isn't great, but he's fine. Yeah? AUDIENCE: How would you
define the consciousness in this context? NANCY KANWISHER: Oh,
that's a good question. And it's a big question. And it's a question that
nobody knows how to answer, not just me. So Christof Koch,
who does more work on the neural basis
of consciousness than just about
anybody, has been going around saying,
for about 15 years, we must not get stuck on
a premature definition of consciousness
because we don't know what that thing is that
we're trying to understand. So I'll hide behind Christof's
parry of that question and say we'll talk about
it later in the course. But there are many
different ways of defining it from the
difference between being awake versus asleep, which is
some of the functions that go on here, the difference
between being knocked out and completely unconscious
under general anesthesia, which is different
from being asleep. Those kind of states
of consciousness are regulated, in
part, in here, yeah. OK, so you can get by
without a cerebellum. But it's not recommended. Moving right along, all
those subcortical bits, we're just going to talk about three
of the most important ones, the thalamus, this
big guy right smack in the middle of the brain,
very large structure, the hippocampus,
and the amygdala. OK, let's talk
about the thalamus. Think about the thalamus
as a Grand Central Station of the brain, OK,
with all of these connections going to all those
parts of cortex coming in and out of
the thalamus like that. OK? So one of the key things
about the thalamus is that most of the
incoming sensory information goes by way of the thalamus
en route to the cortex. OK? So if you start with
your ear, there's sensory endings in
your ear that we'll talk about later in the term. And they send neurons into
this, the thalamus here, this yellow thing, through
a bunch of different stages. They make a stop
in the thalamus. And then they come up
here to this green patch, which is auditory cortex. OK? Similarly,
somatosensory endings, touch sensors in
your skin that enable you to feel when
you're being touched come in through the skin. And they make a stop
in the thalamus. And then they go up to
somatosensory cortex up there. OK? Similarly, visual signals
that come in from your eyes make a stop in the thalamus and
then go up to visual cortex. OK, what's the name of the
structure in the thalamus that those axons
make a synapse in? Coming up from the eyes,
you make a synapse here. And you go up to visual cortex. AUDIENCE: LGN. NANCY KANWISHER: LGN, perfect. What does it stand for? AUDIENCE: Lateral
geniculate nucleus. NANCY KANWISHER: Perfect. OK, you should know that. This is review from 900, 901. OK, yes? Sorry. OK, which sensory modality does
not go through the thalamus en route to cortex
between the sensory nerve endings and the cortex? Sorry? AUDIENCE: Olfactory. NANCY KANWISHER: Yes. Yes. You guys are on the ball. Yes, olfactory system is
the one sensory modality that doesn't make a
stop in the cortex. You can sort of see that here. From the nose, it goes straight
up into olfactory cortex right there. All right, so that's
the standard view of the thalamus is this
kind of like relay station where all the external sensory
information comes in there, makes a stop, and then
goes up to cortex. OK? That's my thalamus act. Boom. Like that, right? OK. But, increasingly,
there's evidence that the thalamus is much
more than a relay station. And why would you bother
with a relay anyway? Kind of doesn't mean anything. Kind of means like
we don't know what's going on here because
you wouldn't just make a synapse for no reason, right? OK, and so the
first thing to note, is there are lots
of connections that go back down the other way? There are 10 times
as many connections that go from primary
visual cortex right here in me, right
here in this guy in red, there are 10 times as many
that go backwards down to the thalamus as go forwards. That's mind blowing, right? Information comes from the
eyes up into the brain. What the hell are those things
doing going backwards, OK? Well, they're doing all
kinds of interesting things. So that's the first indication
that the thalamus isn't just relaying stuff in a
stupid, passive way. And the second
whole line of work, which many people
are working on, but I think some of the most
awesome work on this topic is done by our own Mike
Halassa in this department. And he does these
incredible studies that you can do in mice with
these spectacular methods that we can't use in humans,
where he can really take apart the circuit and
magnificent detail. And he's showing
that the thalamus is involved in all kinds
of high-level cognitive computations in mice. It's really stunning work. When the mice have to switch
from doing one task to another, the thalamus plays a key role in
gating the flow of information from one cortical
region to another, OK? All right, moving along,
the hippocampus, I you guys all learned about this. The number one gripe
in this department as we learn about
H.M. in every course. So that's going to happen here. But it's going to
last about 20 seconds. So here goes. That's a normal slice
of the brain like this. Here's the hippocampus
on either side. It's like a whole curled up deal
right there and right there. And here is H.M.'s brain,
the famous H.M., who had surgery to remove his
hippocampus on both sides, and completely lost his episodic
memory for anything that happened after his surgery. OK? You all remember that, right? If anybody hasn't heard
of H.M., send me an email. And I'll give you some
background reading. OK, so very loosely,
the hippocampus involved both in this kind
of long-term episodic memory that H.M. lost. And it also plays a
key role in navigation, which we'll talk about in
great detail in a few weeks. And I just want to say
that some cases are even more extreme than H.M. So there's a case of
Lonni Sue Johnson. And I am trying to
get you guys a video. And I didn't get it in time. But I'll show it to you later in
the term if you're interested. Lonni Sue Johnson
had a viral infection that went up into her brain. She was an extremely
accomplished person. She did illustrations on
the cover of The New Yorker. She was a pilot. She had her own
farm in which she raised lots of stuff, a
very smart, interesting, multitalented woman, who
had this terrible tragedy of getting viral
encephalitis at I don't know what
age, but middle age. And she now does not remember
a single event in her life. She's smart. She's funny. Her personality
is totally intact. She can answer questions. She can paint. She can do all kinds of things. But she does not remember
a single event in her life. That's pretty astonishing. Reflect on what it means
to have the sense of self if you don't remember
anything in your life. Yeah? AUDIENCE: Can she
remember her name? NANCY KANWISHER:
That's a good question. I'm not sure she. Might know her-- yes,
she does know her name. Actually, it is
evident in this video. But the video, well, so
she doesn't remember. At one point in this
video, she's asked, were you ever married? And she's lovely and sweet and
gentle and kind of low key. And she's like, you know,
just don't remember. I might have been. I might have been. She was married for 10 years. So that's the hippocampus. Important. You don't want to lose that one. Yeah? AUDIENCE: About H.M.,
if the hippocampus is used in long-term memory,
why is it that it being removed caused him to not form memories? NANCY KANWISHER: Well, so
long-term memory means-- it's a vague term. It means the formation
and retrieval of memories that are
going to last a long time. So in H.M.'s case, he can
access a lot of the memories from before his injury. In Lonni Sue's case,
she can't do even that. OK? All right, the
amygdala, OK, amygdala is a Greek word
that means almond. Because the amygdala is the
size and shape of an almond. And so just for fun, we're
passing around some almonds, my favorite kind. Have some almonds
and pass them around. All right, OK, so
the amygdala is involved in experiencing
and recognizing emotions, especially fear. The simple statement that you
should remember about what the amygdala does is just
remember the four F's. You guys all know about the
four F's, fighting, fleeing, feeding, and mating. OK, patient SM lost her
amygdala on both sides. OK? She cannot experience fear. She doesn't recognize
fear on facial expressions of other people. And she doesn't
experience fear herself. OK? And so that's the
striking piece of evidence on what the amygdala does. Her face recognition is
normal, recognizing identities. Her IQ is normal. She's overly trusting
of other people. OK? OK, so that's all
you need to know about the amygdala for now. OK, let's talk about white
matter, just brief review. Here's a kind of tunnel
through a piece of cortex. OK, so my brain cortex is
wrapping around like that. If we took a piece like
this, just took a segment out like that, this is the
outside of the brain up there. Cortex runs like this. And gray matter is the
stuff on the outer surface that's full of cell bodies, OK? White matter are the
axons, the processes that come out of
those cell bodies and travel elsewhere
in the brain. OK? Everybody clear on that? OK, so we got gray matter up
here and white matter down there, mostly myelinated axons
that have that layer of fat to make them conduct fast. And so you'll see bundles of
white matter in the dissection. And so here's an
actual photograph of the slice through a brain. So all that white stuff
up there is white matter. OK, and so you might
say, well, that's just a big bunch of wires. Who cares about that? That's a good question. But actually, the wires
are pretty damn interesting and pretty fundamental. And so I'll just give
you a few reasons. And you don't need to
memorize every one of these. I'm trying to give you a gist
of why we might care about this. And then there will be a whole
other lecture on networks and connectivity
later in the course. Well, first of all, white matter
is 45% of the human brain, OK? So it takes up a lot of space,
all those wires connecting one bit to another bit. And I would say we cannot
possibly understand the cortex and how it works or any little
piece of it without knowing the connectivity of each
piece to each other bit of the cortex, right? Imagine trying to understand
a computer or a circuit without being able to see the
connections between the bits. Like it would drive you crazy. That's the situation we're
in now in human cognitive neuroscience. It, frankly, drives me insane. But that's where we are. Next thing, the
long-range connectivity of each little bit of cortex,
some little bit right there in my brain, is
connected to some bunch of other remote
regions in my brain. And that particular
set of connections is distinctive for
that patch of cortex. So you can think of it as
a connectivity fingerprint of a patch of cortex. OK, so one of the ways that
the different bits differ from each other is by
way of their connectivity fingerprints. And I'm going to skip
the rest of these because we're going to
get back to them later. And I'm going to
run out of time. And I'm going to assign the
TAs to sound the gong at 12:15. OK? Good. All right, now we're
up to the cortex. This is really,
laughably, shallow. But whatever, that's
what we're doing here. So here's this cortex. And as I mentioned,
it's a whole big sheet. And the different bits look
really similar if you just look at them or slice them up. So how are we
going to figure out how this thing is organized? Well, OK, now we're up
here talking about cortex. All right, let's start
with the easy parts, which you've already seen. You've already
seen this up here. These colored bits, visual
cortex, auditory cortex, somatosensory cortex, gustatory
taste cortex, those bits are like the easy
parts of cortex. Those are called
primary sensory regions. There's also motor cortex right
in front of sensory cortex. So those are the
primary regions. They're primary in
the sense of this is the first place that
sensory information lands up at the cortex coming up
from the senses, right? OK, and all of that input is
wired through what structure? AUDIENCE: Thalamus. NANCY KANWISHER: Yes. Thank you. So how are these
regions organized? Well, they have maps. Every one of these
regions has a map. And each of them has a
map of a different thing. So let's start
with visual cortex, and we're going to
talk about the map that lives in visual cortex. But the prior condition
for understanding that map is to understand the concept
of receptive field, which you should know. So I'm going to whip
through it quickly. OK, so here is how
you map the receptive field as a property of an
individual cell in a brain. OK? So the classic way in
animal neuroscience is you place an
electrode in the brain next to a neuron in
monkey visual cortex. OK? So here's this monkey. He's got an electrode
right in his brain right next to a neuron
in visual cortex. And every time that neuron
fires, you get a spike. You hear a spike. OK, now you train the monkey
to stare at a fixation spot without moving its eyes. OK, I can do this with
humans without training you. I can just tell you, look
at the tip of my nose. OK, so keep your eyes
on the tip of my nose. I can see if you're
looking elsewhere. So look at the tip of my nose. OK? OK, so you train a
monkey to do that. That takes a few months. And then they can do that. And then while recording
from neurons in his brain, you put stimuli over here,
put a flash over there or a flash over here or a flash
over here or a flash over here. OK, you can stop
looking at my nose. It's not all that fabulous
a nose, I realize. OK, so a receptive field is
the place in the visual world that makes a given neuron fire. OK? So if there's a
neuron in your brain that responds to a flash here
but not a flash here or here or here or here, the
receptive field of that neuron is right there. Everybody got that idea? OK, so in visual
cortex, neurons have restricted receptive fields. They don't respond to anything
anywhere in the visual field. They respond to a
particular place in space. OK, if that's confusing
at all, ask a question. Because it will come
up again and again. All right, so that's what
the rest of this slide says, what I just said. Blah, blah, blah. It doesn't matter. That's a receptive field. Different visual neurons have
different receptive fields for different parts of space. Now here comes the
important idea. In visual cortex,
two neurons that are next to each
other in visual cortex have nearby receptive fields. OK? So that's the
concept of retinotopy or the map in visual cortex. So you basically have a
map of the visual world in your visual cortex because
there's this systematic layout just like you have
in your retina. In your retina, visual
information comes in. And because of optics,
different parts of your retina respond to different
parts of the image. But that information is
propagated back through the LGN up to primary visual
cortex where you still have a map of the visual space
up in primary visual cortex. OK? So that map is called
retinotopic in visual cortex because it's oriented
like the retina. And so here's a particularly
kind of gruesome but very literal depiction of this
property of retinotopy in a monkey brain. This is an experiment done
very long ago by Roger Tootell. And what he did was he used
a method called deoxyglucose. And so what deoxyglucose
is a molecule that's a whole lot like glucose. But it's got one little
change in the molecule, which means it gets stuck
on the metabolic chain. And so it gets taken up by cells
that want to take up glucose. And then it gets stuck in
there and can't be broken down. So it builds up in cells that
are metabolically active. OK? So you can put a
little radioactive tracer on deoxyglucose, inject
it into a person or an animal. And what happens is it builds
up with this radioactive tag on all the cells
that were active. Make sense? OK, so Tootell did an
experiment where he had the monkey fixate on a spot. And he presented
this stimulus here. So the monkey's
fixating right there. And this stimulus is
flashing on and off. He injects the radioactive
deoxyglucose into the monkey while the monkey's
looking at this. And then, I'm sorry to say, he
killed the monkey, rolled out visual cortex into a sheet. And there it is. And you can see the
bullseye pattern that the monkey was looking
at across the surface of visual cortex. Does everybody get that? OK, so that shows
you very literally what a retinotopic
map is in the brain. It's just like the map of the
visual world in the retina. But there it is up in
the back of the brain. And humans have this too. OK? And so this can be shown in
humans with functional MRI. We'll talk later more about
the methods of functional MRI. But here's a very
high-resolution functional MRI experiment done by some
people over MGH Charlestown. By the way, when I
have names on slides, it's just because, in science,
we don't get paid that much. And so our credit for our cool
data is kind of all we have. And so I can't stand to
talk about other people's cool experiments without
giving them credit. I do not expect you
to learn the names. It's just my little
personal tic that I need to have their name
there to give them credit, even though you don't
know who they are. OK. OK, so what this guy
John Polimeni did was show human subjects
this stimulus here. They were fixating right there. And the stimulus is
flickering with the dots kind of dancing around. And then he looked on
the back in visual cortex on the surface of the brain,
and he sees an M there. It's the same stimulus. It's just flipped
upside down, which is not deep or interesting. The cortex has to be
oriented one way or another. The brain doesn't care whether
you turn it around, right? And your map of visual
space is upside down in the back of the head. And you see that M. Does
everybody get how that also shows retinotopic
properties in the brain in human visual cortex? OK. All right, so the key
idea of retinotopy is that adjacent parts
of the visual field are mapped to adjacent
parts of the cortex. All right, OK, a little bit of
terminology just because people are fast and loose
with these things. I've already referred to V1
and primary visual cortex. It's also sometimes
called striate cortex. It's all the same thing. It's the part of
the visual cortex where the information first
comes up from the LGN right back here. So in me, it's right there. Most of it is in the space
between the two hemispheres. But a little bit
sticks out on the side. So in this person, that
yellowy orange stuff, that's primary
visual cortex, which is the same as V1
and striate cortex. OK? That's just terminology. All right, just as we have
maps for visual space, we have maps for touch space. And so you've probably
seen this diagram here of the map of touch space going
across somatosensory cortex like this. So this is a picture
of a slice like that, showing you which parts of the
body are mapped out to which parts of space. And you can see that
particularly important parts of the body get
bigger bits of cortex. Yeah? OK, just as we have visual
maps and touch maps, we have auditory maps
in auditory cortex, which is right on the top of
the temporal lobe right in here. And what's mapped out
in auditory cortex is auditory frequency,
high versus low versus high
frequencies of sound. And so you see that here's a
piece of auditory cortex in one subject, showing
you regions that respond to high frequencies, low
frequencies, high frequencies. Here it is another
subject, high, low, high, another subject,
high, low, high. OK, so the point of all of this
is that primary somatosensory cortex has maps. Everybody clear on this? The different sensory modalities
map different dimensions. OK, so what about
the rest of cortex? Like you can see,
most of the cortex is not primary sensory cortex. Is the rest of cortex just mush? Or are there separate bits
like primary sensory areas? And if so, do those
other bits have maps? And if so, what
are those maps of? OK? We just took you
from 100 years ago to the cutting edge of the field
is asking this question in lots of different ways right now. OK? OK, let's back up and ask,
what counts as a cortical area anyway? I just posited that these
primary sensory regions count as distinct things. They're like the things, right? They're separate
things in the brain. OK? And if for no other reason,
then they get direct input from the thalamus, right? OK, but let's back up
and ask, what exactly is a cortical area? And we're going to
consider this question by considering the three
key criteria for what counts as a cortical area. OK, the first one is that
that region of cortex is distinct from its
neighbors in function. Neurons there fire in
response to something different from the neurons
in the neighboring region. OK, that's very vague right now. But we'll illustrate that. The next one is-- I mentioned this before--
each distinct region of cortex has a different
set of connections to other parts of the brain. It has a distinct
connectivity fingerprint. OK? And the third thing is,
for at least some regions of the cortex, they're
physically different. If you slice them up and stain
them and look at them really carefully, they
might look a little different than other
bits of the cortex. OK? So those are three of
the key criteria that have been used to say, this bit
of cortex, it's a thing, right? It's distinct, right? OK, so let's look at
the classic example beyond those primary regions. Those are the most
classic regions. Those are the primary regions
we've already talked about. Those are the ones nobody
would fight you on that. This one is next in line. Nobody would fight
you if you say, visual area MT, that's an area. Well, they might. But most people wouldn't. OK, and then from there on out,
it's all fighting all the time. OK, so let's talk
about visual area MT. It's a little patch of the
cortex in a monkey brain. This is a side view
of a monkey brain. And in this human brain, it's
that little patch right there. OK, so this region
meets all the criteria to be a distinct visual area. So how do we know this? Well, we know this from lots
and lots of different methods. So I'm going to whip through
a few of those to give you a gist of how we
can find evidence that that region is
distinct in functional connectivity and the physical
stuff, sometimes called cytoarchitecture. OK? All right, function,
how would we know that region has
a different function? Well, one way,
the classic way is to record from individual
neurons in monkey brains. So if you stick a neuron
into monkey visual cortex while the monkey is
looking at the stimulus that I'll show you
in a second, you'll hear the responses of
an individual neuron. Each click will be the response
of an individual neuron to the stimulus. So let's play this thing, except
it's not making any sound. Chris, can you help me? Oh, right. Duh. That part, OK, see when the
bar of light moves this way, it makes a lot of firing and
not when it moves the other way? Let's watch it for a second. Watch the bar move again. See? It responds less
when it's moving in a different direction. Everybody got that? What is this area
right there called? Yeah, this area right
here in the middle. AUDIENCE: [INAUDIBLE] NANCY KANWISHER: Exactly. That's the receptive field. That's the part of visual space
that makes this neuron fire. OK, this neuron also has a
property called direction. It's sensitive to
motion, as you see. But it's also specific to
specific directions of motion. Everybody see that? OK, so that's a
direction-selective neuron in monkey area MT. And here's a way of showing,
with data, what you guys just saw. This is a map of
different directions in polar coordinates. And this shows you how much-- this is a single cell
being described here. This is the direction
selectivity of that cell, showing you that when
the stimulus moves in this direction, you
get a lot of firing. When it moves in this
direction, you get less firing. And can everybody see how this
plot shows you the direction selectivity of that cell? Make sense? Right. OK, so that shows you what
you just saw in the movie. So this is one way to establish
the function of visual area MT is stick electrodes in
there and record directly from them when a monkey looks
at different kinds of stimuli. And you see direction
selectivity when you do that. OK, further, if you actually
do this systematically, moving across next door
bits of monkey area MT, what you find is that,
as we said before, nearby bits of cortex respond
to similar things, in this case, to similar directions of motion. So here's a little diagram. As you move across the cortex,
you see a systematic change in the direction
selectivity of neurons as you move across the cortex. So in MT, we have a map
of direction preference, just as we had a map
of spatial location in primary visual cortex. Make sense? OK, now because those neurons
are clustered like that-- I forget what my next point was. No. Never mind. We'll get that in a second. OK, what about humans? OK, so here's a monkey brain. Here's a neuron
in a monkey brain. What about humans? Can we record from
single neurons in humans? What do you think? Do we ever get to do that? Yeah? AUDIENCE: Like neurosurgeons. NANCY KANWISHER: Yeah. Yeah. Neurosurgeons,
very occasionally, enable us to record
from individual neurons in human brains. It's the most awesome data ever. Of course, we only do it
when the neurosurgeons have decided, for
clinical reasons, to put electrodes
in human brains. They need to do this to map
out epilepsy before surgery. And sometimes those
patients are super nice and say, yes, I'll
look at your stimuli or listen to your stimuli while
you record from my neurons. And then we get the
most awesome data ever. But it's very, very rare. I don't know of any
data where people have reported individual
neurons in area MT in humans. Yeah? AUDIENCE: So how
powerful should an fMRI be to be able to record
such information? NANCY KANWISHER: Oh,
we're getting there. OK, so given that
we, very rarely, get to record from
individual neurons in humans and we want to more
generally if there is an MT in humans, what do we do? We pop subjects
in an MRI scanner. And we show them moving
dots or stationary dots. And we scan them
with functional MRI. We'll go through the
details of how this works more in future lectures. But what you see,
basically, is this is a slice through
the brain like this. And you see this region
right here responds more to the moving dots. This is the response. This is time here. This is when the moving
dots are on high response. And then when it switches
to stationary dots, the response drops. OK, so with functional
MRI, you can also find the visual area empty
by the higher response to moving than stationary dots. Does that make
sense, more or less? I mean, I'm not giving
you any of the details. But for now, they
don't really matter. OK, so that's cool. But does that tell us
that neurons in human MT are specific for the
direction of motion? Yes? AUDIENCE: Are the moving dots
moving to a specific location? NANCY KANWISHER: They're
moving in all the directions you see here. No, it doesn't. It tells us it's sensitive
to the presence of motion but not the direction of motion. OK? So if we want to really know,
is human MT like monkey MT or is this really human
MT, we want to know, are the neurons in there not
just responsive to motion but are neurons specific
for particular directions of motion, OK? So how would we do that? OK, well, there's lots
of ways of doing that. But actually, one of
the charming things is you can do that
without an MRI scanner. That is it won't
tell you whether it's MT you're looking at. But we can ask the question
of whether your brains have neurons that are tuned
for particular directions. So for this demo, I want you
to fixate right in the center. And do not move your
eyes from that dot. And I'm going to keep
talking for a while, while you keep fixating
right on that dot. And so what I'm
going to show you is something called
an after effect. This is also known as the
psychophysicist's electrode. Psychophysicists are people
who just measure behavior. And from behavior,
they can infer how individual neurons work. And that is about as
awesome as it gets. That's much more impressive
than just recording from the damn neuron. Inferring from
very indirect data how the neuron works from
behavior, now, that is pretty-- oops. OK, sorry. Look directly at my face. You see anything? I didn't see it stop. OK, we're going
to-- oh, here we go. Oh, right. OK, just fixate on
the center again. Sorry. I forgot this guy
was going to stop. So keep looking at the center. And then when it
stops in a little bit, then keep your eyes
right on that dot. And you can see what happens. AUDIENCE: [INAUDIBLE] NANCY KANWISHER:
Oh, that's right. Good point. Yes, right now,
it's alternating. Nothing's going to happen. But that's OK. We're going to have
the whole experience. Keep fixating on the dot. It's good the TAs
are on the ball. OK, fixate on the dot. Anybody see anything? Not really. That's OK. You're not supposed to. That's the control condition. It was alternating directions. OK? So I think it's going
to start moving again. I'm not sure. Let's go back. Let's just start it again. OK, I'm sorry I blew
it the first time. But let's just get this right. OK, fixate on the center
and just keep your eyes right on that center. So this one, it's
not alternating. And it's going to do this
for around 30 seconds. And so the whole point of
this is a way with behavior to ask the question
of whether you have neurons in your brain
tuned to specific directions of motion. And something as low-tech
and simple as an aftereffect can tell you that. Keep looking. Did you guys see anything? What did you see? What happened? AUDIENCE: It wasn't
moving exactly [INAUDIBLE] NANCY KANWISHER: Uh huh. Well, it actually should-- well, now it's doing
something else. But it should shrink at the end. Did you guys see it shrink? OK, so that's an after effect. And the simple
version of the story is that you are tiring
out your neurons that are sensitive to outward
motion while you stare at all that outward motion. And after you kind of burn
them out and exhaust them, then when you look at
something stationary, it looks like it's going inward. OK? And the general idea is
you have pools of neuron-- the easiest way to
account for that is you have pools of neurons tuned
for different directions. And that's why, if you
tire out one batch, you have a net signal
in the other direction. Does that make sense? This is all very relevant
to your assignment which is due tomorrow night at 6:00. This phenomenon was used in the
scanner for that experiment. You can think about how you
would use this phenomenon to ask whether there's direction
selectivity, not just responses to motion, in human MT. Yeah? AUDIENCE: I'm just a
little bit confused. So even when an image
is completely still, like even if you're
not detecting motion, those neurons are still firing? NANCY KANWISHER:
That's a good question. But most likely,
the simple cases-- this may have not worked
beautifully, in part, because I screwed it up and
didn't notice when it stopped. But if it works well, you should
get a pretty powerful sense that after you see it
expanding, then when it's still, it should seem to
be contracting. So when that happens-- the reading assigned for
today, tomorrow night tells you what
happens in your brain during that time when you are
looking at stationary stimuli but experiencing motion. So there's no motion
in the stimulus. But there's motion
in your percept. OK? So that's the question. All right? So read the paper and find out. Yeah? All right, so all of that tells
us just that there are neurons someplace in your brain that
are sensitive to the direction of motion. It doesn't tell us that
they're in MT in particular. But the assigned reading
will talk about that. OK? Right, a further
bit of evidence is remember I said how, in
monkeys, next door bits in MT have similar
direction selectivity. That means you can also
inject an electrical signal in a little patch of MT and
give the monkey a net percept of a direction of motion. OK? If all the neurons were
scrambled around spatially, so that there was no clustering
of neurons sensitive to, say, this direction of
motion, then stimulation wouldn't do anything. But if you train a
monkey to tell you what direction of
motion he's seeing and you show him just random
dots that aren't moving in any direction and you
stimulate one little patch, it'll tell you the
direction of motion of the neurons in
that little patch. And that is much more
powerful evidence that that region is not
only responsive to motion but causally involved in
your perception of motion. OK? I'm a little obsessed with this
distinction between recording responses and
establishing causality. So we'll go over this
in more detail later. But I want you to start
getting used to that idea. Another way to test the causal
role of area MT in motion is with patients with
brain damage in area MT. So there's one
famous patient who had brain damage
right there, which is right where MT usually is. And she could not see motion. And she reports
all kinds of things like difficulty crossing the
street, difficulty catching balls, difficulty
pouring water into a cup, OK, just as you
guys saw earlier. That's called
akinetopsia, right? Kinetics, motion. A, not motion, right? Opsia, eyes. OK. All right, so I started
with these criteria for what makes something
a distinct area. And one piece of
evidence is function. And I just give
you a whole bunch of different kinds of
evidence for distinct function and visual area MT,
that it's specifically involved in motion processing. And the two other criteria,
which are getting short shrift, but I'll just toss them off. And we'll return to them. One is the distinct
connectivity of that region. OK, so you may have seen
this horrific wiring diagram of visual cortex in monkeys. I think it comes up in like
half the talks in classes in my field. This is the one down here. And so there's lots and lots
of different visual areas. And there's a whole
fancy wiring diagram. And smack in the middle of this
diagram, that's visual area MT. And if you blow this
up and stare at it, you'll see that MT has a
particular set of connections to other visual
regions in cortex. And its particular
set of connections are different from
the connections of any of those other regions. It's part of its connectivity
fingerprint or signature. And that's another piece of
evidence that it's a thing. OK? It's not just another like
amorphous bit of cortex. It's a particular
thing in the brain. And finally, you might
wonder, is that bit of cortex physically different? Are the cells in
there different? Are the layers of cortex
different in any way? And you may remember,
from probably 900, about Brodmann areas. Like this dude
Korbinian Brodmann sliced up lots of dead
brains, looked at them under a microscope,
and argued that there were 52 different
parts just from what it looked like if you slice
them up under a microscope. OK? So we called those
Brodmann areas. And area 17, this
primary visual cortex, comes from Brodmann's
terminology. And so he argued that there-- he thought these were
distinct organs in the brain. And he even inferred the
specific histological differentiation of
the cortical areas proves irrefutably their
specific functional differentiation. Well, it doesn't. But never mind. Kind of sounded good. Anyway, that was his idea. And these kinds
of distinct, kind of cellular, physical,
anatomical differences are very salient for
primary cortical areas for vision and audition
and touch and motor cortex. But they're much muckier
for lots of other areas. One important exception,
which is why we chose this, is area MT. And so I'll end in one minute. But just to tell you
where this is going, this is a flattened piece
of monkey cortex rolled out like with a baking roller. No. I don't know. Something like that. So here's monkey cortex. And there's V1 and V2. And it's a big mess. But that big dark blob,
this bit of cortex is stained with something
called cytochrome oxidase. And that indicates
metabolic activity. MT neurons are very highly
metabolically active. And so here's a map
of visual cortex. And that exactly is area MT. So area MT actually
is histologically or cytoarchitectonically
different from its neighbors and fits all of the criteria
for a cortical area. OK? I went one minute over. I realize I threw out
a lot of terminology. I don't want you to
memorize too much. So I made a list of
the kinds of things that you should understand
from this lecture, the things that I think are important.