Much Of This Material...
...was covered in the book "A Model of the Brain" by John Zachary Young, 1964, and alluded to by me in previous posts.
CNET, April 7, 2010 11:01 AM PDT: "...The human eye and optic nerve pre-process visual information into triangleness, circleness, and other geometric shapes. The serifs on letters increase the amount of information sent to the brain, and thus increase speed and (probably) comprehension."
TheRegister, Posted Thursday 5th August 2010 13:22 GMT: "The Human Eye does not send a bitmap to the brain. The three bitmaps generated by the rods (Y) and two types of cones (I and Q) are enhanced and correlated beginning at the sensors themselves, and further decoded along the optic nerve into values like "circleness" "squareness" "nearness" even "faceness" before they reach the visual areas. Just the kind of information needed to solve folding problems.
Because these analyses are so very compute intense when attempted with linear processing, the slow (~44Hz) but massively parallel human eye/brain has a speed advantage.
It is interesting to note the similarities in the human hearing system, transcoding sound from the pressure domain and transmitting a frequency analysis to the brain."
It makes sense that the memory of images follows these same transformations
On a side note, when you make a .jpg file of random data, for example a cloud of pollin, the compression logic encodes non existant eyeballs, faces, and other organic forms into the image.