Events. Um.

We're going to get started because we're going to try and fit in two
mental meta quizzes, plus the lectures. Um, while those of you who've been here
for a while might have read this. This is a warning to not use artificial intelligence
to help do your homework, because you might have noticed the multitude of
errors that this unnamed, um, chat bot has introduced into a very simple
question about the on and off pathways of cells. Um, if you haven't spotted all
the mistakes, which are many, then um, you won't do very well on the quiz. So
without further ado. We'll switch over to the quiz. Um. I'm joined with the QR
code if you want. I'll give you a few seconds and then I'll start. Because I do
something like four questions. So, um, you can see how you get on, on the right.
Okay. Ready? So. This is the easy ones. Minimum number of photons needed to
stimulate a rod. To hold on a single photon is sufficient to give some millivolts.
Notice this is. Which statement is not derived. Yeah. So the correct answer is this
one. This is a classic exam mistake to say that rods can't encode polygons or any
sensitive to one wavelength. This is not true. Rods are sensitive for range of
wavelengths, but the whole population will show the same wavelength
sensitivity. So you can't do this comparison of activation levels for determining
colour. Can I get a bit harder now? I'm not expecting 100% right answers for. This
one. Oh, good. Okay. 29 if you did. Yeah. Absolutely. Key. So facial receptors in
response to increases in light intensity and decreased glutamate release. That's
okay. Um, yeah. So this is the right answer. Um, remember off bipolar cells are
going to their preferred stimulus is a decrease in light intensity. So then they will
hyper polarise in response and increase in light intensity. Answer two probably
the most difficult question. Okay. Um, it's lucky that we're going to I think that's
the end of quiz one. See if you. All right. Well done to these people. So, um. This
leads us to the second lecture, and we're going to, um, start. You'll be pleased to
know by recapping the other pathways very briefly, um, and then expand on how
they used to construct the receptive field properties of ganglion cells. This will
lead us to a second example of lateral information flow. And after that the data
by horizontal cells which you haven't met yet. And then in the second half of the
lecture, we're going to leave the AI and follow the reference pupil pathway to
primary visual cortex. So to recap what we learned this morning, we saw the
retina was a layered structure. We looked at this direct vertical pathway from
facial receptors to bipolar to ganglion cells. We saw that actually in the case of
the rod system it's not quite direct. Right. The rod bipolar connects for an African
cell and ganglion cells. And we looked at the subdivision of both in the rod
system and the cone system of the on and off pathways. So as a quick reminder
we're not going to we're just going to focus on the system for the weather this
afternoon. Um, remember in the cone system that the critical synapse that
distinguishes on and off responses as the spur synapse between the
photoreceptor and the bipolar cell and its sign inverting in the case of the on
pathway mediated by this receptor. And we are six and it's time I'm conserving.
In the case of the octopus, where the cells are stressed and perceptive. So now
we're going to try to understand what these other pathways are useful for. And in
order to appreciate this, we have to learn a little bit more about the response
properties of ganglion cells, which are a bit more complicated than we've
considered so far. So in order to understand the receptive fields of the sorry, the
response, which is gang of cells, you need to introduce a key concept, the
receptive field, which you have perhaps already met. Um, this term was coined
by Hartline in 1948. He was recording from, um, accent and the optic nerve of a
frog. And what he found was that if he shone a little spot of light, like the laser
pointer on small patches of the retina, his axon corresponding to one ganglion
cell would respond, but it wouldn't respond to the sunlight in other regions of the
retina. So he termed the um Coined the phrase receptive field to describe this
little patch of retina, which obviously corresponds to a small region of visual
space to which the ganglion cells responded. And in general, we can define a
receptive field as the region in space in which the presence of a stimulus light in
Heartland's case affects the electrical activity of the neurone. The a word that
seeing a lot of the work in this definition is this word space. So in heartlands
case, this corresponds to a portion of the retinal surface, which is a region of
visual space. But in other sensory systems it can be used in a rather different
way. So in somatosensory system, space might be part of the surface of the skin,
which is a sensory neurone is sensitive to stimuli. In auditory system it might be
part of the auditory frequency spectrum, and the cells are described as having a
receptive field in some position in the auditory frequency space, so be aware that
this term, this word can be used in quite flexible way, depending on what
sensory system uh Is under study. But in the visual system, we usually talk about
the spatial receptive fields of neurones, meaning a region of the retinal surface
which corresponds to a portion of the visual field. So now we know what
receptive field is. What's interesting about the ganglion cells. Receptive fields. So
Steven Kotler um, is working at around the same time as Hartline also studying
ganglion cells, receptive fields. And he found, like Hartline, that individual
ganglion cells would respond when he shone light on a restricted part of the
retina. But he additionally discovered that the way that the cell responded
depended on where he shone the light within the receptive field. So, in other
words, receptive fields have some sort of internal structure, and as a result of his
work and others, it was discovered that the receptive fields of ganglion cells can
be just can be divided into regions which are called the centre and the surround
that are circular. Concentric regions, a central region and this sort of penumbra,
the surround and the centre. And this round almost always have opposite on and
off response properties, which means you can end up with two main types of
ganglion cell, the so-called on centre off surround cells or off centre on surround
cells. So this is all figured out by shining a spotlight on the retina. And here you
can see some of the data supporting this claim. So I found that if a little spot of
light in the centre of an on centre cell. The cell will, as you know, respond to this
increment in light intensity by increasing its firing rate, a typical on response. So
this is the basically the spiking of the cell. And this little line is when the stimulus
is presented. And you can see in response with the spot of light a little increase
in firing rate. If he shone a spot into the office around this on centre of the cell,
then you get an off response. So the firing rate suppressed. Um, if he increases
the size of the stimulus in either the centre or the surround, those responses are
accentuated. So you either get a stronger response or a more impressive off
response. And if he's if he just bathes light across the entire receptive field, then
you see very little response in the cell at all, which tells you that these on and off
surrounds act antagonistic. And so the full name for these receptive fields is the
centre surround, antagonistic receptive field, where the on and off have opposite
effects on the response of the cell. Um, for an off centre cell everything is
reversed. So if you shine light in the centre, you get an off response or
suppression. And if you shine light in the surround, the on surround, you get a
response. This elevation of firing rate. So what we can see from this data is the
ganglion cells really like it. There's differences in light intensity between their
centre and their surround. And they're not very impressed by continuous light or
dark across the entire receptive fields. And this means you'll probably get a
sense already. They're very good at detecting spatial contrast. Remember we
introduced the idea of spatial contrast in the first lecture of changing light
intensity across space ganglion cells because of this receptive field organisation
are very sensitive to spatial contrast. And that's illustrated here, um, where we're
imagining, um, a stimulus, a very simple stimulus in edge. It might be the, you
know, the edge of a door frame or something and cast onto the retina. And we're
going to consider what the response would be of five on centre cells that have
the receptive fields in different locations, um, with different parts of the centre
and the surround seeing the dark part and the light part. So, um, as you can see,
cells A and E are the condition that ganglion cells aren't very responsive to that a
C is entirely dark and entirely light across this receptive field. And these are not
conditions that stimulate the ganglion cells very well because we're activating
both the antagonistic centre and the surrounds simultaneously. So you can see
on this lower plot, this is supposed to be baseline firing here in red. And you can
see cell A hasn't changed at all. And Sally has just modulated up a little bit. So
not much of a response. However cell D is in a in a much more optimal position.
It has got its entire on the centre in the light, which will increase the firing rate of
the cell and part of its office around sees darkness. This decrease in firing rate
from the shadow. So both of those things contribute to an increased response of
cell D. So you can see it modulates positively quite strongly. As kind of the
opposite. The UN response sees a D, sees a decrease in light intensity, and part
of the office around sees an increase in light intensity. Two things that are both
going to suppress the response of the cell. And it shows a negative modulation.
And C is another of these symmetrical cases. Here we have equal and opposite
activation of the centre and the surrounds which cancel to give the full response.
So as you can see as a population, the ganglion cells are pretty good at detecting
edges and visual scenes. And this is emphasised perhaps even better by this
slide, which shows the simulation of the responses of a large population of
ganglion cells to a naturalistic image. So on the left here is the image that is
going to be cast onto the Julian bunny. This is going to be projected onto the
retina. And on the right we're looking at a simulation of the response of ganglion
cells that have their receptive fields. Um, but all these different locations, each
pixel a different receptive field centre. And in this image a sort of grey level
represents not very much modulation. White represents an elevated response
from on centre cells, and black will be an increase in response from off centre
cells. And hopefully what you can see is that the ganglion cell population are
very good at accentuating edges in the image. So if we look at this very striking
edge here, this t shirt, whatever it is, you can see that on the left hand side
there's a strong positive modulation of the on centre cells there on the side to
see the bright t shirt and part of their off surrounds see this dark background. So
that increases the firing rate. And on the other side the off centre cells are
activated. Their off centres see the dark background and a little bit of their own
surround sees the t shirt. So this positive modulates the of the centre cells. So
this is a really nice example I think of how um ganglion cells as a result of this
centre surround receptor field organisation detect a visual feature which is
spatial contrast. And we will return to um frequency tuning inside time. So now
you have a sense of the, um, how the receptive field is organised with this, these
circular concentric centre and surround regions. We're going to try and see how
they're assembled by retinal circuits. So remember this receptive field is a
portion of the retinal surface to which the ganglion cell, you know, through the
retinal circuitry is responsive. So it should be pretty obvious to you that if the
ganglion cell is responsive to events happening in the centre and surround it
somehow, going to have to get information from the photoreceptors. These
circles are supposed to be photoreceptors, the photoreceptors that occupy the
central region and the photoreceptors that live in surround. Both of them have to
somehow provide information to the ganglion cell. How does this how is
accomplished? So the answer lies in another example of lateral information flow
in the retina. And that's mediated by a new interneuron that we're going to meet
in this lecture. The horizontal cell here it is. So it has a cell body in a nuclear
layer. And it makes contact with photoreceptors in the outer flexible layer. And
the important thing about this horizontal cell is it mediates a process called
lateral inhibition. So its job is going to be to collect input from the appropriate
photoreceptors that live in this, around these. You know, the difference in colour
is not very striking. These darker grey things are supposed to be the surround
picture receptors. The horizontal cell is going to get input from those
photoreceptors. And it's going to inhibit the presynaptic terminals of the
photoreceptors that live in the centre. And the result of this, as we will see, is
that the response of the bipolar cell is maximised either in positive or negative
direction. When the centre surrounds, the opposite changes in light intensity. So
whenever there's opposite increments in light intensity in the centre in this
round, the response of bipolar cell will be accentuated and we will follow through
the logic of why that is the case. Um, in almost the next slide I think. so this is
another, um, uh uh, diagram to emphasise this point that the we now in the first
lecture we saw the, the, um, direct vertical pathway through the retina from
photoreceptors onto bipolar cells and then onto ganglion cells. But we now
discover that bipolar cells actually access another source of information to these
photoreceptors that live in the surround by means of the horizontal cell, also feed
into the bipolar response. So there's two routes a direct route from the central
photoreceptors. This can be actually just one photoreceptor if it's a midget
ganglion cell in the in the phobia. Or it can be many thousands in peripheral
retina. I'm here. I guess it's somewhere in between. But they they have a direct
path to the bipolar cell. And then the surround photoreceptors have this indirect
route through this lateral information flow pathway mediated by the horizontal
cell. So why does this configuration, um, result in these antagonistic receptive
fields? So, um, right, this is the this is the slide where you need to concentrate
the most in the lecture. So we're going to consider the response of an on centre
of surround cell. That's the receptive field organisation to remind you. And we're
going to imagine giving this cell its preferred stimulus which is going to be a light
increment in the centre like that and a light decrement in the surround. Okay. So
let's start with the centre, the easy one. So if we show a light increment in the
centre we know what's going to happen. The photoreceptors will hyper polarise
and release less glutamate. The because this is an on bipolar um it will
depolarise it's expressing it will depolarise. And as a result it will release
glutamate onto the ganglion cell and that will increase its firing. So so far so easy
right. That's a normal on pathway. Um, but now in addition to this, we're going to
give a light decrement in the surround. So if we give a decrease in illumination in
the surround, these photoreceptors are going to depolarise and release more
glutamate. Activating the horizontal cell. And the horizontal cell in turn will
inhibit the presynaptic terminals of the centre photoreceptors. Remember, these
terminals are already hyper polarised because these cells saw their increment in
light intensity. But now they're getting additional inhibition from the horizontal
cell, which is going to hyper polarise even more and further decrease glutamate
release, which means the response of the bipolar cell will be even stronger. And
as a result of the positive modulation, that ganglion cells will be even stronger.
Yeah, it looks like. Mostly get it? So the you know, you should be able to use this
type of logic to think through what will happen if we also show an increase in
light intensity in the surrounded the cell, or to think through the response
properties of an off centre cell? My things will be reversed. Um, the key things I
guess to remember are that both the ganglion cell and the bipolar cell show this
central sound receptor field organisation, because they get input from reception
centre directly. And this surround the photoreceptors is fed in through the
horizontal cell. And because this is a an inhibitory connection, you can also think
of it as sine inverting the input from the surrounding photoreceptors. Okay. So
um, this is an outline of basically how the centres around receptive fields are
constructed for the ganglion cells. Um, uh, one thing to appreciate is there's a
broad diversity of different types of retinal ganglion cells, which we're not going
to unpack in great depth. Um, but I'm going to briefly mention some of the
properties of the two major classes of ganglion cell in primates. So in primates,
for a long time, people have described how cellula or midget ganglion cells on
the one hand and magna cellular or power cell ganglion cells on the other. And
as you can probably guess from the name, these cells differ substantially in size
and of the cell itself and the size of its receptive field. So the the power of
cellular cells are small. Um, they can have also the rates of bondage in primates.
About 90% of the ganglion cells in the primate retina are part of the cellular cells.
Um, they're particularly abundant in the central retina. And the macula and the
fovea and their receptive fields can be my need. So these are the cells where the
centre can just be a single cone in this area. And then they have a surround of a
few more cones. So they have very high visual acuity. Their receptive fields can
just be a tiny fraction of a degree in width, which provides the limits on your
visual acuity. And we'll come to some of their other properties in subsequent
slides. The parallel cells or cellular cells are much larger. So these cells are found
in the peripheral retina. The receptive field is is built up from input from, you
know, thousands of photoreceptors. And um, what else is there to say about this?
They, uh, well, they, they often show transient responses to events that happen
in their receptive fields and are thought to be involved in motion processing,
which is something we're not going to get into in any depth. But these are the
main distinctions that you see sometimes. Um, one interesting thing about the
power of cellular cells is that some of them are sensitive to differences in the
wavelength of light in the centre and the surround, and these are the famous
colour potency cells. So there are two types of colour potency red versus green
and blue versus And, um. Um. This is an example of a a red on centre green
office around cell. So this cell receives input in its centre from Elkins. So an
increase in red light intensity in the centre will excite the cell. But this round
primarily gets input from cones. And so a decrease in green light intensity in the
surround will also activate this cell. Um, if you put white light across the
receptive field, the cell doesn't respond very much. And two versus yellow is the
other system where the centres have bias cones. And the surround yellow gets a
mixture of M and alpha in input. So as you might imagine, these cells are very
well placed for comparing the level of activation of the different cone types,
which, as we discussed in the first lecture, is the basis for colour perception. In
the cellular system, there is no such colour potency because these cells receive a
mix of different photoreceptor inputs and both the centre and the surround. And
so the the system is colour-blind and as I mentioned, more interested in motion
processing. So there's all sorts of different types of ganglion cell. And we've seen
some of the ways in which they vary. Um, one important aspect of how they're
arranged in the retina is that the different, the functionally distinct ganglion cell
populations form so-called mosaics across the retinal surface. And a mosaic is
kind of illustrated in this slide. The idea is that for a particular population of
ganglion cells, they're receptive fields. And the cells and cells neatly tile the
retina, but with very minimal overlap, as illustrated here, for each different type
of ganglion cell forms its own mosaic. So the on parasols form their own mosaic,
the on midgets form their mosaic midgets and etc. and these independent
overlapping mosaics of different classes of ganglion cells contribute to the
parallel processing that we discussed in the first lecture. So if you remember, he
says, that there's a range of different spatial frequencies, in contrast, that we
want to capture from visual scenes. And we imagined this scene being encoded.
Now hopefully you can understand how this is achieved. So if we imagine this
image projected onto the retina, ganglion cells have large receptive fields will be
responsible for encoding a low spatial frequency representation of the scene.
But, um, the cells are very small receptive fields, but the centre and the surround
are very tiny. And so they're sensitive to contrast modulations over very short
length scales. In other words, high frequencies. This population will be the one
that captures the high spatial frequency representation of the scene. And these
things are happening simultaneously. These independent mosaics are doing their
own processing with the visual input to generate different what we call these
spatial frequency tuned representations, which can be communicated to the
brain to support behaviour. So. This leads us then to the brain. So we're going to
now in the second part of this lecture go out of the retina um to the brain. So as
we've the retinal ganglion cells we've just been talking about are the projection
neurones of the eye. So they're extending axons through the optic nerve to
different retinal recipient brain regions. And there are a number of different
regions in the brain that get input from directly from retinal ganglion cells. And a
few of them are listed here. So the superior colliculus is located in the midbrain
and supports all sorts of interesting aspects of midbrain vision. It's particularly
important in orienting and avoidance responses in all sorts of vertebrate species,
and it controls psychotic eye movements, um, also projection of the projection,
which does different things, including the capillary reflex to change the size of
capillary of the pupil in response to changes in light level. And in primates, at
least a majority of the ganglion cells project to the lateral geniculate nucleus and
the thalamus, which is, um, a relay station. It's probably an unfair way of
describing it en route to visual cortex. And this pathway from retina to LGM, uh,
to cortex is the one that we're going to look at. Um, now in sort of support
conscious visual perception. So, um, let's have a look at this. Um, right. A
pathway to the cortex. Um, you might already be familiar with it. So the ganglion
cells project, um, out of the ion along the optic nerve, and the axons, um, pass to
the optic some of the midline of the brain. And at this point, some of the retinal
ganglion cells cross the midline and others state ipsilateral. So the rule is that
ganglion cells from the nasal side of the retina, this is nasal relative to the fovea
which we're going to treat as zero degrees. These are the ganglion cells whose
axons cross and those on the temporal and the temporal retina. So that's the
side of the retina here relative to the phobia based ipsilateral. And as you can
see hopefully from this slide the consequence of this is that the entire right
visual field. So the whole visual field on the right hand side of your nose comes
to be represented in the left hemisphere and vice versa. The left visual field is
going to get represented in the right side of the brain after crossing or not at the
optical zoom. The cells carry on in the in what's known as the object track, until
they reach the lateral geniculate nucleus in the thalamus. This is where the
ganglion cells make their make a form a synapse, and then the postsynaptic
neurones, um, they project the rest of the way up to primary visual cortex. The
lateral geniculate nucleus in the thalamus is a histological section. Have a look
at. It is another layered structure. Um, and it has um a bunch of different layers,
uh, in which these different sort of, uh, feature channels we've been discussing.
Input from the left and the right eye are kept separate. So you can see there's
different layers receiving left and right I input. And there's also segregation of the
cellular and the cellular input from the cellular ganglion cells and the magnet
cells themselves. This is also kept separate. What about the receptive field
properties of the neurones. So there's a very coarse approximation there similar
to retinal ganglion cells. So they also have centres around antagonistic receptive
fields. And those in the public health system will have small receptive fields that
might show colour repellency. Those in the magnet layers will have bigger
receptive fields and won't show they'll be colour-blind but motion sensitive. So to
see a real changes in receptive field properties, we have to wait until we get to
the visual cortex, which is now. So here is the primary visual cortex and the
occipital lobe at the back of the brain. It's also known as V1 or striatal cortex or
area 17. They're all the same brain region. We'll call it V1. And this is the primary
site that's going to receive ascending input from the thalamus, from the LGM
part from V1. Um, uh, information passes out into other regions of cortex. It's
thought there's about 20 different cortical regions that contribute in some way to
visual processing as an approximation. Um, and in your textbook, you'll see two
main routes described for the flow of information. So one is the so-called dorsal
pathway from V1 into the uh, parietal lobe, uh, which is thought to be particularly
important for our processing, uh, space and movement. So it's sometimes
described as the rare pathway, but there's also a ventral pathway towards the
temporal lobe, which is thought to be more important in object recognition and
sometimes called the the what pathway. Um, in this lecture, though, we're just
going to focus on the response properties and uh, different aspects of V1 primary
visual cortex. So cortex like the lion and the retina is another layered uh,
structure. This is a histological section looking through the layers of the cortex.
And as you probably know, in neocortex, um, six main layers are described. Um,
these are sending inputs from the thalamus, the neurones which are going to
provide the, the, the the relay of the visual input from retina. Um, they, they
predominantly in primates innervate layer for. So this is the main retinal recipient
layer. But information can pass from layer four uh, to other regions of cortex via
so-called intra cortical connections in which there's two main types. So the ones
that span across the different layers of the cortex are called radial connections.
So that will take visual information from the airport to for example, layer two
three. And there are also horizontal connections in the plane of the cortex which
can take information from V1, for example, out to V2 or MT, other cortical
regions, and on to the what on the where pathways. So let's have a look at
what's going on in layer four. So a little bit like we said the left and right eye
information is segregated in the LN. Uh, this segregation is retained actually in
layer four. And this was shown by, um, a very famous experiment done by
Hubble and Wiesel in the 70s. And what they did in this experiment was they
injected a radioactive tracer. Radio labels proline into one eye of an animal. This
radio label's amino acid was taken up into proteins, uh, in the retinal ganglion
cells and transported along their axons to the LGN. And then it could jump
across, amazingly, jump across the synapse in the end to label the entire retinal
fugal pathway all the way from the injected eye into, uh, into, um, the what and
what you're looking at here is an auto radiograph that you've done these little
tours looking down on layer four where this sort of it's not very clear in the
projector, this white um, uh, signal represents the Activity from these acts on
terminals that derive eventually from the injected I, o and the black spaces in
between represent input from the um injected I. And these um bands are the
famous ocular dominance columns, so there are about half a millimetre wide,
and they represent the segregation of external inputs from the LG n the
representing left and right on information. So this is shown in the 3D. Here the
left and right eye information is kept separate. And therefore. And as a result of
this postsynaptic neurones and their form monocular, they only respond to
events happening in one eye or the other. But in other cortical layers like layer
two three, you do find lots of binocular neurones, and that by popularity is
established by these cortical connections. So connections between the layers
mean the neurones in theory, for example, can respond to events in both eyes,
which as you can imagine, is important in binocular vision. So this is the first
example of several. We're going to look at of how functional information is
organised in V1. So these are the ocular dominance columns that are found in
layer four. Another organisational feature is retinal topic mapping. So you might
have heard the term retinal before. What it means is that adjacent regions on the
retina, which correspond to adjacent regions in visual space, are going to be
represented by adjacent neurones in some target structure. So there's a rich
topic map in the cortex which we're going to look at. But there's also retinal
maps in several other brain regions too, like the LGR that's also retinal map and
the spherical nucleus in the midbrain. Another classic example. So how what's
the evidence for this? This is one piece of evidence. So this is a visual stimulus a
sort of wagon wheel thing that a monkey was viewing. And then the, um, uh,
activity in V1 of the monkey was visualised using radioactive glucose. Another
radioactivity experiment, um, to, uh, label metabolically active cells. And then
the monkey, um, donated its, uh, v1 study. And here you can see the pattern of
activity across the cortical surface, the vote by this visual stimulus. And it's
pretty obvious that the this pattern is reproduced across the cortical surface,
indicating that at least that there's rather gross scale adjacent regions in visual
space, or activating neurones that sit next to one another in cortex. And of
course, this retina, this retina arises from a very orderly pattern of connectivity
between the retina, uh, to the LGM and then to the cortex or adjacent retinal
ganglion cells. We're going to pass some information eventually to adjacent
neurones, um, in cortex. So as I said, this is a sort of level of organisation, a quite
a zoomed out, um, large scale. Um, you're probably wondering, so, um, about
the response properties of individual neurones in V1. So Hubel and Vsel again
discovered that their receptive fields are quite unlike the receptive fields we've
seen in the retina, ganglion cells and in the um ln neurones. So um, remember
those neurones had a circular concentric organisation with a centre and a
surround, um Hubel and discovered that in V1 neurones also had on and off
subregions. But they didn't show this circular concentric organisation. Instead
they were elongated regions lying side by side. So let's have a look at some of
their data from the 50s and 60s. Um, this is data from the cat. So they were
showing the cat, um, little spots of light and mapping the receptive fields of
individual cortical neurones. Here's five examples in this video. The little crosses
represent on the responses and the triangles are off responses. And if we have a
look at this cell for example let's say it's on C. You can see it has this long
elongated on subregion flanked by two off surrounds. So you get like your sort of
three fingers right. An elongated on subregion and two offs around either side.
Um, the overall size of this receptive field, to give you a sense, is about 4 to 12
degrees in width in the cat. But these this little on subregion could still have a
fairly narrow width on the order of about one degree. So it's about the width of
your index finger at arm's length. So does still capture detail. Um, one um, uh,
very interesting property of these styles, which you can probably start to guess
from this receptive field organisation, Is that there? Um, they really like it when
you present them with oriented bars at a particular orientation. So this is
illustrated here actually for an example cell. So this cell prefers vertically
oriented bars. And you can see if people in Vsel projected a vertical bar onto the
retina. This cell showed a very powerful burst of action potentials. But as the
orientation of the bar deviates from vertical, you can see the rate of firing
decreases eventually to zero when it's, um, horizontal. And this preference, uh,
um, for edges or bars of a particular orientation, hopefully can be understood
simply by looking at the structure of the receptive field. So let's have a look at
this cell E. It has the central off region elongated region flanked by two on
surrounds. If you put a dark edge across the receptive field of this orientation,
that's going to provide a strong response from the off centre and, uh, from these
flanking on surrounds, but if the bar is to cross the receptive field. But an
orthogonal orientation like this, you'll have equal and opposite activation of the
centre and surround, which are antagonistic and very little response on the cell.
So these cells, by virtue of this receptive field organisation, really like oriented
edges. Um. Oh yeah. This morning I kind of thought it was funny. Uh, maybe
when you're revising visual neuroscience to think of an analogy with various
breakfast foods. So, um, the retinal ganglion cells are neurones a bit like fried
egg. They have a circle of concentric on and off subregions, the centre in this
round, whereas the southern big one a little bit like three sausages with these
elongated, uh, on and off subregions lying side by side. Um, some of you might
find that useful and revision. Um, don't worry if not. So this property of
orientation selectivity in any event, and pseudoscience is something that has
been studied a lot in cortex and is another feature which is neatly organised in
the cortex. So Hubel and Vsel found that if they sink an electrode down through
the layers of cortex, they find orientation selective cells and those cells along
this trajectory down through the depth of the cortex all had the same orientation
tuning. So in this example, they're all tuned to an orientation of let's call it 45
degrees. Um, except in layer four you don't get orientation selective cells. This is
so-called a famous orientation column down through the depth of the cortex. But
if they move their electrode the other way across the plane of the cortex, they
encounter cells that show progressively different orientation tuning. So you can
think of the cortex being made up of a whole bunch of columns of cells, where in
each column the orientation tuning is slightly different. Um, this has also been
studied using an optical imaging method called intrinsic imaging. So this is the
method where you look at the surface of the cortex and assess the spectrum of
reflected light in order to infer changes in blood oxygenation. So it's a kind of low
resolution method for kind of mapping functional activity on the cortical surface.
It's still used by people who do mammalian neuroscience. And here's some
example data. So each of these little images shows one of the activity maps that
you get if you show a bar of a particular orientation sweeping across visual space
to activate V1 neurones, fire receptive fields at different locations. And white
indicates activation of the of these cells. So the colour code these different maps
and put them together. You can generate a composite like this. And what's
observed is that there are patches of the of the cortex where there are bunches
of cells that have the same orientation preference. So this little green patch here
are all cells which like it are oriented like this. And the red patch all cells with a
horizontally oriented bar. So these are called ISO orientation domains. These
patches, and they represent essentially a cross-section through those orientation
columns that are going down to the cortex. So going down into the cortex is
going into the board looking down on the on the surface of cortex in this case. So
orientation is another thing that is nicely organising cortex. So Hubel and Vsel
did a lot of study of response properties of cells in D1. And um this slide is to
illustrate there's a little bit more complexity beyond what I've described. So
many cells in V1 are tuned to orientation. They like bars of particular orientations
as you've seen. Um, but uh, Hubel and Vsel found that there's at least two kind
of broad populations, um, depending on the organisation of their receptive fields.
So the simple cells are the ones I've been describing so far. They have receptive
fields that have clearly defined on and off subregions. Range, like the sausages
lying side by side is on subregion flanking off surround. But they also identify
cells called complex cells, which although they also like oriented bars at a
particular orientation, you can't find clearly defined on and off subregions inside
the receptive field. Instead, they could actually evoke on and off responses
throughout the receptive fields. And um, one hypothesis is that their receptive
field properties are somehow assembled by combining input from simple cells.
There's also a type of cell called the hyper complex cell, which we won't talk
about. So, um, Hubel and Vsel also hypothesised how this receptive field
organisation might be assembled. And it's still a hypothesis to this day. Um,
remember that the input to the cortex is coming from the LG N and in the LN the
cells kind of have receptive field properties like retinal ganglion cells, the fried
egg. So they have circle concentric centres around receptive fields. And they're
not orientation tuned. So orientation selectivity is somehow an emergent
property of V1. And Hubel and Basil proposed that perhaps it was generated by a
kind of wiring diagram like this. So they imagined this cortical simple cell getting
input from a bunch of neurones whose receptive field centres are neatly
organised. And the little row sees in this case the on centre cells and they in turn
getting input from retinal ganglion cells with the same kind of configuration, a
little array of ganglion cells with um uh on centres neatly lined up across the
retinal surface. And they thought that if this, uh, is the case, that perhaps these
on centres can superimpose to generate this long sausage like um on subregion.
And then the officer rounds can also superimpose to account for these flanking
officer rounds. Um, they also thought maybe you could have something even
more complicated where there might be a mix of inputs from both on and off
centre cells, which are offset in space. So here, these off centre cells, with their
off centres slightly displaced, they could superimpose to also contribute to the
office around in a slightly more complicated, uh, configuration. Um, in any case, I
think we still don't know how the response properties of the V1 simple cells are
generated. And this is, uh, to this day, the best proposal. So one final. Yeah, one
final functional property which we're going to have a look at in cortex is direction
selectivity. So we've said that cells prefer a bar in a particular orientation moving
through the sorry percentage of that receptive field. Some cells are also
interested in the direction in which an edge moves through the receptive field. So
this animation, which I made myself, is a receptive field of a V1 cell that prefers
vertically oriented bars. And this cell responds really well if the bar moves from
left to right through the receptive field. That's this increase in spiking here. That's
called the preferred direction, but it responds much less well if the bar moves in
the opposite direction, the so-called null direction. And this property property is
called direction selectivity. The cell is selective. The left or right motion in this
case. And as you can imagine, it's yet another functional property which is not
haphazard across the cortex, but sort of neatly organised. And here's some
evidence of that again from intrinsic imaging. So here a what was it. I think it's a
ferret is being shown um, oriented balls sweeping in different directions across
space. These activity images are being, generated and combined to generate an
activity map, and you can see that cells that show the same directional tuning
are clustered together into patches on the cortical surface. So we've talked about
a number of different functional properties. Um, how they also come together in
terms of cortical organisation is sort of cartoon. And here is a picture from uh,
one of your textbooks of what's called a, um, uh, cortical module. Yes. So the
idea of the cortical module is it's like a little block of cortex because of the rat
and topic mapping, all the neurones in this little block have their receptive fields
in the same place. They're all doing the same part of visual space, but they have
different types of tuning properties. So, um, in the in layer four, the segregation
of information from the left and right eyes and the dominant columns throughout
the layers of the cortex, except in layer four we have these orientation columns
cells responsive to different orientation of edges. They also show directional
selectivity, which is not illustrated in this cartoon. And these things, the blobs are
something we haven't talked about. They thought to contribute to colour
processing, but the the kind of take home idea is that this cortical module is
little, um, element that views a particular region in space, has a rich
representation of the sort of colour and orientation and other properties of edges
that exist in that part of space which support, um, uh, visual perception and
visually guided behaviour. So this is obviously a simplified cartoon. Remember,
the real data looks more like this. So here again is this image from intrinsic
imaging showing these ISO orientation domains. We have said that some cells
also show direction selectivity. And the way this is organised by the way is within
an ISO orientation domain. There's a scene called a fracture which separates the
cells that are attuned to movement of that oriented edge in opposite directions.
So here, for example, on the left hand side of the neurones that like a bar of 45
degrees, moving kind of that way. And on this side, the cells that like movement
in the opposite directions, this fracture separates these populations. I'm down in
Liverpool. We have the dominance columns. And then this whole structure is
repeated across the surface of the cortex according to the anthropic map. So. We
might have time for the mental EEG quiz. Let's see. So this is the summary of
this lecture. We hopefully you've understood how the receptive fields of retinal
ganglion cells are composed so that they're responsive to contrast differences in
light intensity across our receptive fields at particular spatial frequencies. And
how this depends on lateral information flow from the horizontal side. And the
second part we've looked at various aspects of organisation of response
properties in primary visual cortex, and seen that in almost all cases, they're
neatly organised and refused to slide over the cortical margin. Um, right. We
have maybe we have just about one time later invested to do the next quiz. Um,
and you can access as. You said. Yeah, if you want to. Do you want to do the quiz.
Stay and do it if you need. Thank you. Okay. Hopefully everyone has joined to
respond to this. Okay. First question. It's about the construction of the central
receptor field ganglion cells. This requires some. Construction. As well. Okay. Way
too much time. Okay. All right. Well, I guess that's good news. Uh, 21, if you got it
right. So remember. Yeah, inhibition from the horizontal cell will increase When
the horizontal cell gets more input from feature sectors, and that happens when
you have a decrease in light intensity that causes the cells to release more
neurotransmitter. Right. So the light intensity, more glutamate release on the
receptor and therefore activation of the horizontal cell, which is going to then
inhibit. The central receptors. Okay. Good. Some of you will do well. Okay. Read
the question carefully. The typo in this answer the two of you chose that one.
Okay. This is correct. The the reason the face that the country is the high acuity
is because the ganglion cells have small receptive fields. The centre can just be
fed by a single cone, the five ganglion cells. All. Right. So my breakfast. Analogy
like these. Some of you weren't paying attention to the breakfast. Yeah. So the
concentric centre surrounds receptive fields is retinal ganglion cells and
neurones. Right. But the simple cell in cortex and V1 has elongated on a lot of
subregions. right? And because it's a simple cell, we can tell which are the own
regions in which of your regions. If it was a complex cell, we wouldn't be able to.
So that might be the small. 100% right answer to this question. Oh, nice. Well
done. Yeah. Everything on top of that. Directly selected neurones, columns and.
Location. Very good. I didn't realise there's another question. Okay. What's this?
One of Off bipolar cells. Huh? Okay, back to lecture one. See if we can do better
on the bipolar ones. This is the question the majority of what you might think. So
see if we can do better this time. Oh. Yeah. Maybe if you sleep on it, you will. The
information will consolidate in your brain. So important to have the consolidation.
Yeah. Remember depolarisation is preferred. The small cell is a decrease in light
intensity. So.