Okay, so I've been too long from this, too long from noding in general, but here you go, the first of the second set of updates in How your brain works.

The Retina

Okay, first I want to mention here, that by discussing the retina, I'm again stuck in the world of vertebrates. Such is the nature of the neuroscientific community, unfortunately. I want to point out that invertebrate eyes, and visual systems in general differ greatly from vertebrate systems. For example, we cannot directly observe the polarization of light, while most invertebrates with enough complexity of sight to have a visual system can. I think, perhaps that some fish can detect light polarization, but I'm not entirely sure. Regardless, it's a cool feature that we lack, probably for a multitude of reasons. But I want you to think about this for a second. We are missing out on a whole other component of light that the rest of the world is in on. We think of light as having wavelength (color), and intensity, in our day to day experiences, but we totally forget about polarization. It's like another dimension that we're missing out on. If you don't think it is significant, do some research on it, and see if you can find pictures in which they show an original, and one in which the light polarization has been made artifically apparent. It's a whole other level of detail, something entirely out of our range of perception, and it's nearly impossible for us to imagine what a percept of such a thing would be like. It would be like every object you saw having (possibly) two colors at once, on top of one another, but not mixing. Freaky eh? Just adds a bit to the beauty and mystery of life. Anyway, on to the retina...

So the retina is pretty slick. It's one of those interesting things in science, that's just been latched onto by so many people that we know a lot about it, while other topics remain all dark and murky. Kinda like the genetic code for Drosophila. It's just been studied a lot, and thus is overrepresented in the writings and mind of many a neuroscientist, which I'm not quite sure is a good thing. But enough of my philosophical ramblings. Basic function. Right, so light comes through the lens of the eye, and is focused onto the retina. Now, the retina has cells which are sensitive to light, commonly called rods and cones. Okay, most of you have probably heard this before, but basically, cones do color vision, and rods do light intensity (black and white).

Rods: Rods are my favorite of the two. They are sensitive to extremely low levels of light, and it has been shown that (in frogs at least) dark-adapted rods respond measurably to being struck by a single photon. These cells are much more active in the dark than the cones. Notice that when it is really really dark, you don't see colors very well. This is because your cones aren't picking up light anymore, but your rods are. Rockin.

Cones: Cones are typically reffered to as being one of three types: red, green, or blue sensitive. The color labels aren't really all that accurate, because their peak responses occur closer to yellow, green, and violet, but whatever. What matters is that each type responds best to a specific wavelength of light, but also responds to a range of other wavelengths, to lesser degrees. You can graph something that looks like the magnitude of response vs. the wavelength of light and get a 'tuning curve'. The tuning curves of these three classes, like the tuning curves for many sensory neurons, overlap. The overlap is what allows us to see a wide range of colors, with accuracy. Sensory neurons that use overlapping tuning curves can use vector coding to create hyperacuity. In a short definition, hyperacuity is what you get when you use the signals many low resolution input devices to reconstruct detail at a higher resolution than any individual receptor is capable of. I'll try to get into this with some later nodes on neural computation.

Alright, I thought you should know about the two types of sensory structures in the retina. Now that it's done, we need to talk about the really good stuff, the overall strucutre of the retina. Now, here's something that might be a surprise if you've only learned about this stuff from high school bio. There is more to the retina that just rods and cones. Much, much more. There two other layers of cells in the retina, besides the layer made up of the rods and cones. That makes three layers. Count them with me...

One cell layer, HA HA HA (rods + cones)
Two cell layers, HA HA HA (horizontal, bipolar, and amacrine cells)
Three cell layers, HA HA HA (parvo and magnocellular cells)

Three cell layers.
A bit of weirdness about these layers. They're organized backwards from what you'd expect, which is to say that the rods and cones are at the back of the retina. Light passes through the other two layers before reacing them. This is no big deal, really. The layers are pretty thin, and anybody who's tried to look at unstained animal cells under a microscope will tell you - they're not exactly what one would call opaque. The interesting bit about this is that all of the axons of the third layer leave the eye at the same place. To do this, they have to pass through the other two layers, and since the axons are occupying space in the layers at that point, there's no room for rods and cones. So you do have a blind spot, but you don't notice it because your eyes are always moving, you have two eyes with blind spots at different points in the visual field, and because your brain corrects for such things.
Basically they work like this. The rods and the cones detect the light. You geeks out there can think of them loosely like pixels if you like. There is a region of the eye in which the cones are more densely packed, called the fovea. The fovea is the part of the retina which can see in detail. It comprises the portion of your visual field which you use for reading, and closely examining objects. If you want to know just how much of your visual field it takes up, hold you arm out in front of you. Way out. Look at your thumbnail. That's about the visual area in which you can see fine detail. You seem to percieve more detail, because you have the aid of memory, and almost constant eye movement. This eye movement is important. Let's try an experiment. Go get something red. Something primary-type red, with a nice solid color to it, not something having a bunch of different hues and shades of red. And something with nice crisp edges. You don't have anything red? Well get something blue. Okay, blue. You've got something blue, right? Okay, now put it on a white background, and pick a spot close to the edge of it. Stare at that spot, don't move your eyes. Now sing the Jeopardy song twice (60 seconds). Okay, look away and blink your eyes a bit. You should see a yellowish afterimage in the shape of the object. This afterimage is cause by the blue cells that were previously reporting on that image being less active. They have acclimated, and respond less than their neighbors (the cells that were seeing white earlier) which are not acclimated to blue. Less blue gets interpreted as more yellow, because the red and green cones both respond some to yellow light, but the blue cones don't. Neat eh? You might wonder why the cells do this, this acclimation. Well, it's actually an advantage. The problem of afterimages is easily overcome, and acclimation to a static stimulus is important in the nervous system in general. Think about something for a minute. How much of what occurs in you life that is important is static? How often is the mere presence of something vitally important to you. Think about it for a bit, and you'll realize that's what's really important are the things that change, in space, or more frequently in time. That car in the street. Is it important? That depends on if it's moving toward you, or just sitting there, doesn't it? Who cares what kind of car it is if it's about to run you over? You don't need fine details to see it coming and get out of the way. This is why, by the way, your fovea is so small. Most of the eye is devoted to detecting movement. It's an evolutionary thing. Nobody ever got et by a Tiger that wasn't moving. Most of the nervous system's approach to the world is that changing things are important, unchanging things aren't. Even as you read this, what is important to your brain is not the color of the text, or the color of the background, but the change from one to the other, the edges which demarcate the shapes of the letters, the words. Thus, we see the pattern of acclimation in the nervous system. When a sensory modality is presented with something new, it responds strongly at first. Then, as time passes without a change, it responds less and less. You get used to the feeling of having clothes on your body. You hardly notice your glasses anymore. You get the idea.

A quick experimental aside to demonstrate just how good your peripherial vision is at detecting motion. This will only work if you're sitting at a (somewhat older) CRT. Laptop users, it'll also work with a TV, but it's best done with an image that is relatively motionless. Okay, start by looking at your screen. Nothing's moving, right? If something is, make it stop, because it's going to mask what you're about to see. Now, turn your eyes away from the screen, but keep your attention focused on it. When it gets way into the periphery of your vision, hold it there for a minute. You should start to see a flickering. That's your screen refreshing. (Know those black bars you always see when TVs and computers are filmed on home video?) If you have a brand new monitor with a super-high refresh rate, this might not work. (I'm not positive, it's been a hell of a long time since I could afford one of those) I know it works on TV's and most other low end CRT displays though. Cool, huh? Straight on, you can't see it, but from the side... So sometimes it's not always best to look directly act something you're interested in. This is why if you get in a fight, you shouldn't watch your opponent's hands or feet. You should focus on their upper torso, and let your peripherial vision pick up the motion of their attack, as it's more sensitive to such things. Just in case. Or you could just not get in so many fights, you damn rebel.

Of course, rods and cones individually can't detect changes in things. They just present absolute information. Information on environmental changes is hidden in the information, but it's not explicit. The second layer of the retina is where the really cool stuff starts to happen. The horizontal, bipolar, and amacrine cells create lateral interactions between the information from the rods and cones. These interactions, which can be thought of as calculations performed on the initial image, are where our retina gets its heightened sensitivty to edges, contrast changes, motion. This is where the system becomes adaptive. Adaptivity rules. This is why retinas don't work like cameras. Cameras are good for detecting the absolute charactersitics of a scene. Retinas are good if you want to do things like object detection and recongition, in multiple environments, with varying light levels, or even colors of illumination. Oh, how adaptivity rocks. You don't even know yet, but you will....
See, the thing about cells in the second layer is that they don't fire action potentials. Neither do the rods or the cones, actually. This is a nice bit of engery conservation, but it makes them hard to study directly, in vivo. The third layer, on the other hand...

The third layer is the layer of cells which outputs to the optic nerve, which carries the visual information to the brain. There are two categories of cells in this layer, referred to in primates as parvocellular and magnocellular, or P and M. These cells are easy to study, and have certain response properties which can give a lot of insight into what those second layer cell functions are all about. So, I'm going to go through how these cells behave. It's probably the simplest way to get a sense of what's going on in the retina.

First off, you need to understand the idea of a receptive field. The receptive field of a neuron is essentially the region in stimulus-space that the neuron responds to. I say stimulus-space, because a receptive field can include characteristics like color and frequency, not just position.

Arguably the most important type of spatial interaction in the retina is something called center-surround. This is a property displayed by the parvocellular and magnocellular cells. These neurons have receptive fields which are small with respect to the visual field as a whole. They will respond to a stimulus anywhere in this receptive field, but they respond best if the center of the receptive field is lit (or appropriately colored), and the surrounding area is dark (or has a different appropriate color), or vice versa. If the entire region is lit (or... you get the picture), the response is somewhere in between. Here's another of my bad ascii diagrams:






stimulus type		cell type     firing pattern(underlined is during stimulus)
*=lit #=shaded |=action potential

 ____________
|            |
|            |          on-center     |  |  |  | | | | |  |  |  |
|    ****    |		
|    ****    |          off-center    |  |  |  |   |   |  |  |  |
|    ****    |
|            |
|____________|



 ____________
|            |
|            |          on-center     |  |  |  |   |   |  |  |  |
|    ####    |		
|    ####    |          off-center    |  |  |  | | | | |  |  |  |
|    ####    |
|            |
|____________|



 ____________
|############|
|############|          on-center     |  |  |  |||||||||  |  |  |
|####****####|		
|####****####|          off-center    |  |  |             |  |  |
|####****####|
|############|
|############|



 ____________
|************|
|************|          on-center     |  |  |             |  |  |
|****####****|		
|****####****|          off-center    |  |  |  |||||||||  |  |  |
|****####****|
|************|
|************|



Anyway, you get the basic idea. Keep in mind that these are idealized neurons, and not real data. Real biology is rarely that simple or clean. Real neurons do things like acclimate, respond transiently when the stimulus is switched on or off, and other such things. As for the mechanism for creating center surround, it's not hard to dream up a basic neural network that explains the response pretty well. We'll take center-on. You have a small field of receptors (rods, cones, or both) which all connected to the center surround cell. The cells in the center of this field form excitatory synapses with the center-on cell. Their neighboring cells synapse inhibitorily onto the center-on neuron. Really, it's not quite so simple, but you get the idea. So, what's the point of all this center-surround nonsense? Well, center-surround cells are very good for detecting object edges, and shading of the stimulus. Think about it for a bit. It'll start to make sense, if not now, when we get to the LGN and V1.

Ways in which retinal ganglion cells differ from one another: P and M pathways:
These retinal ganglion cells, which project to the brain are (as I said before) usually divided into two main types, Parvocellular and Magnocellular, or P and M for short. This division has a lot to do with the differences between the fovea and the periphery.
Parvocellular cells have most of their connections around the fovea. They have small receptive fields, which means that as a group they have high spatial resolution. This is, of course, good if you want to detect the shape of an object. They are also very color sensitive, again a property for measuring the details of an object. Parvocellular cells typically respond more or less constantly to a stimulus. That is to say, if you present them with a stimulus which either excites or inhibits them, they respond pretty much the whole time that the stimulus is present. This is good for picking up the static properties of an object.
Magnocellular cells, in contrast to Parvocellular cells have larger receptive fields, and are found more in the periphery of the retina. They are relatively color-insensitive, and have phasic off and phasic on responses to stimuli. This simply means that they respond transiently at the beginning and end of a stimulus presentation, not constanly for the duration of the stimulation. This makes them good for motion detection as a group. As an obect moves across the retina, the magnocellular cells will be active at only the leading and trailing edges, where the individual cells "see" the moving object as a stimulus-on (leading edge), followed by a stimulus-off (trailing edge). Thus, there will be more or less a moving outline of activity at the edges of the object.



So that should give you some general idea about what's going on in the retina. As always, please /msg me if you have any corrections or questions.

Now! Deeper into the brain -> The LGN
Or, back to How your brain works

Ret"i*na (?), n. [NL., from L. rete a net. Cf. Reticule.] Anat.

The delicate membrane by which the back part of the globe of the eye is lined, and in which the fibers of the optic nerve terminate. See Eye.

⇒ The fibers of the optic nerve and the retinal blood vessels spread out upon the front surface of the retina, while the sensory layer (called Jacob's membrane), containing the rods and cones, is on the back side, next the choroid coat.

 

© Webster 1913.

Log in or register to write something here or to contact authors.