Although there are millions of rods and cones in the vertebrate eye, it's necessary to be somewhat selective in the transmission of the signal back out. Some light-sensitive elements will be on and others off at any given moment, obviously, and the brain has to interpret this. The connections between the primary elements, the rods and cones, and the brain, are mediated through the various layers of the retina in such a way that the actual input to the brain is controlled.
Since they're neurons, the rods and cones use neurotransmitters as would be expected. But unlike ordinary neurons, they don't generate an action potential. Rather, the rods and cones, and all of the integrator neurons except the final ones in the chain use electrotonic conduction, i.e. the direct flow of electric current along the membrane. Thus there is a continuous charge flow, not the all-or-nothing response seen in typical neuronal transmission. The release of neurotransmitters at the synapses is graduated in proportion to the amount of current flowing, and the response post-synaptic cell (the first of a series of integrator neuronss) is graded in turn. By use of this mechanism of graded conduction and varying levels of neurotransmitters, the response is proportional to the light intensity falling on the area of the retina.
Rods and cones synapse with integrator neurons in the outer plexiform layer. There are several types of such neurons and the pathway by which the signal is conducted is different for rods and cones. Types of integrators include horizontal cells, whose communications are entirely within the outer plexiform layer, and which connect rods and cones with bipolar cells.
Bipolar cells can receive signals from the rods and cones directly, or via the mediation of a horizontal cell. Their dendrites are in the outer plexiform layer, and their axons extend inward to the inner plexiform layer, where they synapse with other integrators, specifically amacrine cells and ganglion cells.
The amacrine cells mediate signals between bipolar cells, other amacrine cells, and ganglion cells. The ganglion cells are the final element in the chain, and they receive input from either the bipolar cells or the amacrine cells. The simplest and fastest pathway for transmission of a signal is:
And the ganglion cells' bundled axons make up the beginnings of the optic nerve. The connection for rods is different than it is for cones, and it involves four neurons in the chain
Rod > Bipolar cell > Amacrine cell > Ganglion cell
with, again, the synapse with the ganglion cell marking the end of intra-retinal processing and the beginning of transmission of the integrated output into the visual cortex.
The signal routes described above are the simplest possible cases: there is considerable flexibility of "choice" in the pathway, and the actual route of information through the retina is usually a great deal more complicated. Further, the influence of horizontal cells on the ultimate output has not yet been discussed.
Lateral connections between two rods and cones, bipolar cells, etc., are all possible and the number of switches in the system and possible combinations is nearly infinite.
In the diagram at right, which is itself a considerable simplification of the actual situation, it is easy to trace any number of possible pathways a signal can take from the rod or cone to a ganglion cell.
The horizontal cells have an important role: they are there to inhibit the other cells. Horizontal cells always have an inhibitory output, and by virtue of being connected laterally to numerous bipolar and rod/cone cells, they can suppress the generation of information along some pathways and routes, while not affecting others that are adjacent to them. This selective suppression of signal transmission is called lateral inhibition, and its purpose is to increase acuity of vision.
When light falls on the retina, it may illuminate some sensors brightly and the others in the area around it less so. By suppressing the output of the less-illuminated areas, the horizontal cells insure that only the highest-intensity output gets through; and hence contrast and visual definition are improved.
Bipolar cells are also somewhat unusual neurons, in that they can hyperpolarize or depolarize, depending on the type. Thus they can send a "positive" or a "negative" signal via the ganglion cells, and the brain has the capacity to interpret these two levels of signal differently. This mechanism allows for two types of signals to be sent, and it also enhances the lateral inhibition phenomenon.
The ganglion cells are the last neurons in the chain, before the output leaves the eye and goes to the brain. There are several types of ganglion cell, and it's believed each type is used for a different purposes. The W-cell is excited mainly by rods, and its function is to perceive directional movement all over the retina. As an object moves across the field, the w-cells are stimulated and pass the information on via the integration chain.
The X-cells are responsible for most color vision. They receive input from the cones; and they are also the most numerous type of ganglion cell. They probably account for the pinpointing of the image in its precise location on the retina.
The Y-cells are the largest and least numerous type of ganglion cell. They appear to be dedicated to the perception of changes in light intensity and perhaps very rapid movement of an image across the visual field.
Ganglion cells, unlike the other integrator neurons, do have action potentials, and even when unstimulated they fire at a more or less constant rate. It is the changes in light intensity and the shifting of the image over the field of vision that causes a change in the firing rate in ganglion cells. These "on" or "off" responses of individual ganglion cells, which take the form of changes of frequency of firing, are interpreted by the brain as the final output of the system. By switching the ganglion cells on or off as dictated by the integration of the image in the retina, the eye has an amazing ability to detect motion of even very small objects.
The dorsal lateral geniculate nucleus also controls how much of the signal actually gets to the cortex. It has internal inhibitory circuits that can selectively turn individual signals off and regulate exactly which visual information is ultimately passed through to the cortex for processing.
The secondary visual cortex receives signals secondarily: they are transmitted to these areas for analysis with respect to motion, shape, position, etc. via intra-cortical pathways. Different regions of the secondary cortex are responsible for different types of classification and analysis; and depending on the "conclusion" reached, actions can be initiated by motor control areas of the cerebrum.