A. After the retina:

Once the ganglion cell axons leave the retina, they travel through the optic nerve to the optic chiasm, a partial crossing of the axons. At the optic chiasm the left and right visual worlds are separated. After the chiasm, the fibers are called the optic tract. The optic tract wraps around the cerebral peduncles of the midbrain to get to the lateral geniculate nucleus (LGN). The LGN is really a part of the thalamus, and remember that nothing gets up to cortex without synapsing in thalamus first (if the cortex is the boss, the thalamus is an excellent secretary). Almost all of the optic tract axons, therefore, synapse in the LGN. The remaining few branch off to synapse in nuclei of the midbrain: the superior colliculi and the pretectal area. We will come back to these.

B. The lateral geniculate nucleus:

"Geniculate" means knee-shaped, and it is a pretty accurate description of the LGN. No matter how you slice it, the LGN looks like a striped Andy-Capp-hat. The stripes are actually layers, and there should be six of them in most parts of the LGN. Each layer receives inputs from a different eye: 3 layers for the left eye and 3 layers for the right. (Keep in mind, each LGN gets information from 1 hemifield, but 2 eyes.) These layers alternate, so if you were to label the axons from just one eye, you would see alternate stripes labeled, as below.

There is a second aspect of organization in the LGN. The outer 4 layers are composed of small cells, and correspondingly, receive inputs from the small ganglion cells of the retina. These layers are called the parvocellular layers. The magnocellular layers, on the other hand, are composed of large cells and receive their input from large ganglion cells. We will return to the reason for this segregation at the end of the section.

C. On to cortex:

The neurons in the LGN send their axons directly to V1 (primary visual cortex, striate cortex, area 17) via the optic radiations. This highway of visual information courses through the white matter of the temporal and parietal lobes, and can be very vulnerable to strokes. Once the axons reach V1, they terminate primarily in a single sub-layer of cortex.

Remember that all cortical areas in the cerebrum are composed of six basic layers, but that each specific area of cortex may have modifications of layers to best serve its function. It was on the basis of layer appearance and cell types alone that Brodmann first subdivided the cortex into over 50 areas. These areas are known today to correspond to functionally distinct areas - area 17 is primary visual cortex, for example. As you struggle to identify layers 1-6 in any piece of cortex, ponder the fact that this man identified 50 different subtleties of those layers. It's awe inspiring.

The layers of V1 are specialized in one primary way - layer 4 has been expanded into 4 sublayers: 4A, 4B, 4Ca, and 4Cb. Layer 4A is a dark layer, while the deeper 4B is a very pale layer full of myelin. 4B is actually visible without a microscope - it is the line of Gennari, the white stripe that gives V1 its other name, striate cortex. Layer 4C is important because it receives most of the input from the LGN. Due to these specializations, you too can see a transition between Brodmann areas. Follow along the 4B stripe, and you will see it suddenly disappear into a more compact layer 4. This is the transition between areas V1 and V2 (secondary visual cortex, area 18).

D. Ocular dominance stripes:

We have mentioned that the LGN axons enter into layer 4C. However, recall that the LGN is segregated by eye. This separation is maintained as the axons enter the cortex. Each cluster of axons from each LGN layer spreads out in a little column within 4C.

As the signal is transmitted to upper layers of cortex, the information from the two eyes is mixed and binocular vision is created, but here in 4C the two eyes are still entirely separate. Therefore, if you could label the inputs from a single eye, in 4C you would see little pillars of label. If you were to cut tangentially (parallel to the surface) through layer 4C, you would see that all those pillars line up next to each other and form tiger stripes. These are the ocular dominance stripes.

E. What exactly IS cytochrome oxidase?

Cytochrome oxidase is a staining technique which preferentially stains metabolically active neurons. This is a useful tool in looking at ocular dominance columns: if an eye is surgically closed in an experimental animal, a cytochrome oxidase stain will reveal the LGN axons carrying signals from the remaining, active eye. This is how one eye can be selectively stained, as in the picture to the right. However, researchers staining with cytochrome oxidase happened upon another property of V1. In layer 3, certain areas always seemed to stain darkly, regardless of the status of the eyes. Tangential sections through layer 3 showed a pattern similar to leopard spots. These areas of stain were named blobs, and the pale areas around them, interblobs. It has since

been shown that they represent a division of labor within the cortex: specifically, the blobs are more involved in color processing than the interblobs.

F. Receptive fields in V1:

Recall that the receptive fields of both ganglion cells and LGN neurons were center-surround, and that they responded optimally to points of light. Neurons in the cortex, however, respond very poorly to points of light. The optimal stimulus for most cortical neurons turns out to be a bar of light, in a very specific orientation. How did this come about? One hypothesis is that the key is in how the LGN axons converge on the cortical neurons.

Take three LGN neurons responding to adjacent ganglion cells. Their ON-center receptive fields are represented by this array of donuts.

Now hook all three of these LGN neurons up to one cortical neuron, and dictate that all 3 must be excited at once to evoke a response in cortex.

What would be the absolute optimal stimulus to drive the cortical neuron nuts? Three points of light, right?

Three precisely spaced points of light are actually not so likely in nature, but how about a line?

Sure enough, a bar of light will turn on this cortical neuron.

How about a bar oriented 90° to the right? You can see that this stimulus would be too weak for the cortical neuron.

The simplest type of receptive field in the cortex follows this arrangement closely. An optimal stimulus is a bar of light in the center of the receptive field, at some precise orientation. Other connections and convergences in other areas of cortex set up more and more complicated receptive fields, until single cells can be responsive only to the shape of a face. But really all of the information for that face entered your retina as nothing more than... ...a thousand points of light.

G. Division of labor in the visual system:

What we see can be divided into several categories of vision: color, linear pattern, motion, etc. The perception of these different categories requires a wide variety of equipment, and some of the divisions are made as early as the retina. For example, rods see only black and white, and can function in dim light, whereas cones can see all colors but require full light. There are at least two types of ganglion cells as well. There are small ganglion cells that dominate in the fovea: they are color sensitive and are "fine-grained", meaning their receptive fields are small enough that they can pick up a high level of detail. These small cells are called P cells (P for parvo for small). The second type of ganglion cells have large dendritic arrays, and receive information from a wide radius of bipolar cells. They are mostly found in the peripheral retina, are not color-sensitive, and are "coarse-grained" and relatively insensitive to detail. Their main asset is that they are sensitive to motion - you can imagine that due to their width they can track a signal flashing across several bipolar cells. These are the M cells (M for magno ).

These two types of information, motion vs. color and form, are maintained in separate compartments all the way up the visual pathway. They are kept in separate layers in the LGN, enter V1 via separate sublayers (4Ca vs. 4Cb), and after passing through V2, go on to separate areas of associative cortex. In the end, the parietal visual cortical areas (such as MT and PP) end up dealing with motion of objects, navigation through the world, and spatial reasoning (which is essentially moving things around in your head). Temporal visual areas (such as V4 and IT) are involved with the complex perception of patterns and forms as recognizable objects.

In summary, when your Aunt Edna goes streaking by you after unwisely taking a dip in the snake-infested pond out back, your color/form pathway will identify her as Aunt Edna; your motion pathway will tell you which way she went. Any emotional response you attach to this event will be mediated by the amygdala, incidentally.

H. The pupillary light reflex:

Way back at the beginning of this section, there was mention of a few optic tract fibers which bypassed the LGN entirely, traveling instead to the less glamorous but equally essential midbrain. One of their targets in the midbrain is the pretectal area, which mediates the pupillary light reflex. This reflex can be demonstrated by shining a light in one eye; if all is working correctly, both pupils will constrict.

Light enters the retina and from there travels directly to the pretectal area. After synapsing, the information is sent to the Edinger-Westphal nuclei on both sides of the midbrain - this is the crucial step in ensuring that both eyes react to light. The Edinger-Westphal nuclei, via the IIIrd nerve, control the pupillary constrictors that narrow the pupils. Knowledge of all this enables you to test the status of your patient's visual system by shining a light into each eye.

For example, if you test each eye, and no matter where you shine the light, the left pupil constricts and the right one remains dilated, what is your conclusion? There must be a problem with constriction on the right, such as IIIrd nerve damage. BUT, what if shining light into the left eye produces bilateral constriction, and shining light into the right eye produces no constriction? Here the problem must be with the right optic nerve itself, or possibly the right pretectal area. What would happen if you made a cut down the midline of the midbrain, severing right from left?