I heard recently that a scientist has developed a retina prosthesis. So I think this is a question that has an answer.

Does the retina encode information like a Bitmap image or an SVG image?

(A Bitmap is a grid of pixels. An SVG image is vector based, so that it defines a shape and says 'in this region, it's green, in that it's red', so to speak.)

  • $\begingroup$ The eye sends a Bitmap to the brain. The brain interprets it somewhat like a SVG where the brain only focuses on the relevant complex information structures in the image that you as a person chooses to focus on. $\endgroup$ Feb 14, 2015 at 20:16

2 Answers 2


Short answer
The retinal image corresponds more to a bitmap, than a vector-based image

The retina contains a layer of about 100 million photoreceptors that are topographically organized. In other words, each photoreceptor codes one specific pixel in the field of view. In turn, the nerve fibers running from the eye to the brain are also topographically organized in the same way the retina is (retinotopy is maintained up until the higher brain centers and is lost only in the inferotemporal cortex). Hence, yes, basically, but grossly oversimplified I agree with @Krysta that the retinal image more closely corresponds to a bitmap than a vector-based image. Nonetheless, colors, contrasts and edges are coded in the retina.

Here is a picture on retinotopy. Basically a 1:1 topographic representation of the retina can be found in the primary visual visual cortex:
source: University of Leeds and Sydney

Retinal implants are devices that replace the lost photoreceptors in the retina in patients with retinitis pigmentosa or other retinal degenerative diseases. Retinal implants typically consist of a grid of electrodes (60 - about 1500 in commercially available systems) that are placed on (Argus II) or below the retina (alpha-IMS).

Here's a fundus photo of an Argus II implant:

Argus II

The electrodes directly stimulate the surviving neruonal cells in the retina. Retinal implants make use of the retinotopy. An electrode in the center of the retina will elicit a spot of light (a phosphene) in the central fielfd of view, whereas a more peripherally situated electrode will generate phosphenes more eccentrically. Hence, camera images (in the Argus II device) are simply downsampled and projected onto the electrode grid as intensity-coded stimuli. In the alpha-IMS system the intra-ocular part consists of a grid of photosensitive diodes coupled to small amplifiers. Hence, in this device it is, quite literally, a photosensitive chip thrown on the retina. In both approaches, brighter areas in the image receive higher electric currents and result in brighter and/or larger percepts.

Stronks et al. Exp Rev Med Dev;11:23-30

  • 1
    $\begingroup$ Wow, both really high-quality answers. This is really cool, thanks. It seems then that it's a bitmap with some svg edges overlayed? $\endgroup$
    – John
    Feb 12, 2015 at 20:28

Caveat: I know very little about the structure of digital images.

Assuming you mean the biological retina and not the prosthetic one, the Wikipedia page on color opponent process is quite good; basically, the four receptor types respond maximally to different wavelengths, and the ranges overlap slightly, so the visual cortex does the differences between the activation levels to get a complex representation of color.

Opponent processing

In terms of where in the visual field (pixels vs "shapes"), again, retinal ganglion cells have what is called a receptive field--they respond maximally to stuff in the middle and less so to stuff on the periphery (whether "stuff" is light or shadow depends on their exact type). So although the spatial definition of a "pixel" is a bit more complicated in the retina than in a bitmap, and simplifying grossly. . . it's more like a bitmap, I think. There's no edge processing or shape definition at the retinal level--there's no "processing" at all in the retina, just transmission--so vector-type processing doesn't happen until visual cortex.

  • $\begingroup$ There is edge enhancement in the retina, among other neural processing mechanisms. But I agree that when choosing between a vector image and bitmap, the retina is a bitmap, but it is a gross oversimplification. +1 $\endgroup$
    – AliceD
    Feb 12, 2015 at 3:37
  • 1
    $\begingroup$ @ChrisStronks Can you say more about image processing in the retina? I had always understood that there was image/shape/edge transmission, but no information convergence between different sensor types or receptive fields such as happens further downstream. Not sure if we are working off of different definitions of image processing or if you have more information, but very curious! $\endgroup$
    – Krysta
    Feb 12, 2015 at 11:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.