The neuronal inter-connectivity of primate retinas can be likened to Recurrent Neural Networks (RNN). The retina being primarily responsible for visual input of biological beings consist of a multitude of neuron layers sending signals in a feedback manner. Rods and cones are the core elements in detection of light and color intensities. The retina is constructed with the level of complexity that puzzled scientists and researches have revealed that the retina does not only capture temporal image focused by the lens and relay it straight to the visual cortex of the brain for processing.
During the development of infants, part of the brain forms the eye in which the neurons are layered in the retina, with a neural network structure that is far different from the brain's. Therefore processing of information begins in the retina, right after visual information is captured. The question is then "If the brain does the processing of visual information, what does the retina do to these information before sending to the brain ?"
The biological retina potentially perform simple tasks like image denoising or complex object extraction. Most researches in emulating the retina's functionality has concluded to advancements in digital cameras like noise reduction, color constancy, sharpening, motion detection, object extraction, etc. These functionalities have addressed limitations individualistically by applying filters to improve the images captured.
Much has yet to be discovered in order to further understand the full potential of primate retinas. We recreate the retinal neural network, sampling a closest possible primate retinal structure. Visual information processing will be carried out in a manner that is similar if not same as a biological retina. Photons (or pixels in image processing) would travel along the recurrent neural network, resembling a biological primate retinal neural network structure..
It is postulated that a recurrent analogue neural network would best represent the biological retinal model.