I don't post about this a lot, but professionally I've been working as a graphics programmer for many years. Some time before I started this career, I was recommended a seminal work on human visual peception, Vision Science: Photons to Phenomenology, which presents a unified view of the problems of vision drawing from disciplines such as physical optics, cognitive psychology, and neurophysiology. I am just starting to read it now, and, as often happens, kicking myself for not having done so sooner.
One thing that stood out to me in the introductory chapter is how, according to models and experiments current to the time - this was published in 1999, so I am not sure how up to date this view is - the way visual data is analyzed is broken down in the brain in various ways. In particular, shape and position are analyzed in two different parts of the brain, as are color and shape in contrast with distance and dynamics. This is interesting because in computer graphics, often there is a separation between model data (shape) and world position (captured by a model transformation), and there is also a separation between perceived color data and depth data, the latter often also compared to past depth data to drive dynamical calculations.
I wonder if further reading will lead to other such correlations. And whether this information was used by the designers of modern graphics processing units and architectures when deciding how to structure the modern graphics pipeline to begin with.
