I have been considering the neural retina recently after a really amazing seminar I attended. This scientist, Chichilnisky, is attempting to create a more accurate artificial retina by studying the processing that happens on the level of the ganglion cells.
So, the back of the eye has minute and well defined layers that must work together harmoniously for vision to happen. The cells that usually detect the light and turn it into visual signals are the photoreceptors. This vaguely means light receivers. They are supported by the retinal pigmented epithelium, or RPE. Most forms of vision loss that are degenerative result in the loss of the above layers.
This means the eye can signal the brain, but it cannot detect light. An artificial retina, therefore, involves replacing the dead cells with a camera that stimulates electrodes directly contacting the nerve cells that signal the brain. Essentially it creates a new “layer” that can detect light and translate it for the remaining neural retina.
HOWEVER! The way that ganglion cells and other cells process the information that they receive isnt well understood. There are layers of tens of different types of cells that activate in seemingly random patterns in response to visual stimuli. Given that this results in high visual acuity, or good vision, there is clearly a consistent processing algorithm.
The craziest part is that it! Isnt! Evolutionarily preprogrammed! In an experiment on cats, kittens were blindfolded for the first few weeks and as a result were blind. Humans have a similar learning curve - we have functioning eyes and our brains have to learn how to use them. How do they learn? What is the algorithm? How does it work? Is it consistent across people?
All good questions! The speaker basically went into some of the initial experiments and considerations that are needed to design an artificial retina with better acuity than “can see vague shapes and light and dark.”
I really love science and the eyes are amazing organs.