Visual information detected by the retina is sent back to visual centers of the brain for further processing. Special retinal neurons called retinal ganglion cells send long wiry connections, called axons, to visual brain regions via the optic nerve. Most retinal ganglion cell axons cross over, so that neurons in the left retina mainly connect to visual areas on the right, and vice versa. However, a fraction of retinal ganglion cell axons do not cross, and instead connect "ipsilaterally." This arrangement allows the brain to integrate overlapping information coming from the left and the right eye, providing a major cue for mammalian depth perception.
The fraction of retinal ganglion cells which connect ipsilaterally varies from mammal to mammal. There is a fascinating, though imperfect, relationship among mammals between the degree to which the eyes' fields of vision overlap and the fraction of ipsilaterally projecting retinal ganglion cell axons within each retina. In general, species with forward set eyes, in which left and right eyes see more overlapping parts of the visual scene, have a greater percentage of ipsilaterally projecting retinal ganglion cells than animals with eyes set to the sides of the head. This may represent an evolutionary tradeoff between a greater retinal field (in grazing animals favoring panoramic vision) and greater steroscopic vision. Thus primates, with forward facing eyes and extensive reliance on depth perception (think about swinging between branches) have a very tidy separation of ipsilaterally and contralaterally (crossing) projecting retinal ganglion cell axons. Cats, ferrets, and other carnivores have many ipsilaterally projecting cells. Animals like horses, cows, and mice have far fewer. And it is known that sperm whales have essentially no overlap between their left and right eye fields of vision (harpooners used to try to sneak up on them from directly in front) but I don't think anyone has looked at their retinal ganglion cell axons.
During development, the axons of the retinal ganglion cells have to find their way to the correct side of the brain. The decision whether to cross, or to stay ipsilateral, occurs at the optic chiasm , the crossroads where the optic nerves lead to the optic tracts. In the last two years, scientists have figured out quite a lot about how retinal ganglion cells achieve this decision. It turns out that developing retinal ganglion cells whose axons are destined to cross at the optic chiasm express a different genetic program from the ones whose axons are destined to stay ipsilateral. A critical difference lies in two transcription factors (proteins which help control gene expression), Zic2 and Islet-2. Zic2 is expressed by retinal ganglion cells whose axons project ipsilaterally, and islet-2 is expressed by retinal ganglion cells whose axons project contralaterally. These two transcription factors not only control the receptors on the axon which are involved in the decision to cross or not to cross, but they antagonize each other's expression. Knockout mice for either transcription factor have increased numbers of retinal ganglion cells committing to the other decision.
But the coolest of all is that the expression of Zic2 in the retinas of different species reflects the number and location of retinal ganglion cells cells which are known to project ipsilaterally in that species. (Figure 5 in this link ). Mice showed a smaller percentage of these cells than ferrets. But even more intriguing, Zic2 retinal expression also varied in frogs between the panoramic tadpole stage and the partially binocular adults-- corresponding to the development of an ipsilateral retinal ganglion cell connection! Although the brain circuits for binocular vision differ between mammals and amphibians, it seems that these two transcription factors have been involved in the development of binocular vision for a very long time.