Let’s have a look on a grayscale scan of a page from “Through the Looking Glass”, taken at an initial resolution of 300 dpi. We can, of course, read it easily and speedily. Our gaze flows across the page, extracting meaning and old memories. We may concentrate on the pictures, or on page layout, the meaning of sentences and unfamiliar words, the shape of individual letters, morphemes of letters, a stain on the paper. A warped page, bad lighting, crumbled paper, variable viewing angles do not even slow us down. A mirror image, however, is unreadable.
All this happens in one single act called reading.
Our visual system did this, and more, well before Gutenberg’s invention. It has been perfected by evolution during hundreds of millions of years. But how did it all start? Even a primitive eye, bad optics and few sensitive cells, is useful. At the very least, it can warn about a predator that is still far away. The retinal image is small, resolution is low, there is almost nothing to be seen. (Hyperacuity).
Recognition and discrimination evolved from this border case.
If you scroll down to the bottom of the scanned page, you are in for a surprise.
Resolution decreases gradually, until individual letters occupy only a few pixel cells.
The text is still readable!
Even if we use a threshold operation to reduce the image to just two colors, the small italics at the bottom remain readable. ( here )
In the human brain, the small retinal image corresponds to a small patch of cortical tissue that contains a small number of cortical columns. Their output is sufficient for the recognition task.