4 ms·

Interesting stats. Could you elaborate on the data you mentioned please? What does the 40 exabytes represent? Is that a single “frame” of an image captured by t

by cbeach 3y ago

Interesting stats. Could you elaborate on the data you mentioned please? What does the 40 exabytes represent? Is that a single “frame” of an image captured by the eye? Or perhaps its data per second?

stevenwoo 3y ago
We know from studying lower life forms that they absolutely do not send some pixel accurate representation of an image(barring some incredible compression/encoding) From the seminal paper What the Frogs Eye Tells the Frogs Brain https://web.archive.org/web/20110928024235/http://jerome.lettvin.info/lettvin/Jerome/WhatTheFrogsEyeTellsTheFrogsBrain.pdf https://web.archive.org/web/20110928024235/http://jerome.let... to Guinea pig eyes https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1564115 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1564115
tibbydudeza 3y ago
From Quora "Each eye has 137 million rods and cones. If we assume 8 bits per pixel and 60 frames per second for 2.5 billion seconds, then the total information from both eyes is 3.3 x 10^20 bits, or about 40 exabytes (40 million terabytes). More realistically, the retina does not transmit pixels directly to the brain. Instead, it detects low level features such as center-surround contrast and movement. Each optic nerve contains about 1 million fibers. We can't see events lasting less than about 0.1 seconds, so let us assume an information rate of 10 bits per second per nerve fiber. This is a total of 10^6 fibers x 2 optic nerves x 10 bits per second x 2.5 x 10^9 seconds = 5 x 10^16 bits, or 6 petabytes. This represents a lossy compression ratio of 6600 to 1. The brain's visual cortex compresses this data further by detecting a hierarchy of increasingly complex shapes such as edges, lines, angles, characters, shapes, and objects such as faces. In a 1986 study in which people were asked to learn and recall thousands of pictures, Tom Landauer estimated that human long term memory is on the order of 10^9 bits, or 1 bit per second. If we assume that 40% of our long term memory is visual (in proportion to the size of the visual cortex), then it suggests that the retina and brain compress visual data at a rate of (3.3 x 10^20 / 4 x 10^8) = 825 billion to 1." https://qr.ae/pKofqQ https://qr.ae/pKofqQ
- cbeach 3y ago
  Ah so it's the input data over the course of a human lifetime. Thanks!

by cbeach 3y ago

stevenwoo 3y ago

We know from studying lower life forms that they absolutely do not send some pixel accurate representation of an image(barring some incredible compression/encoding) From the seminal paper What the Frogs Eye Tells the Frogs Brain https://web.archive.org/web/20110928024235/http://jerome.lettvin.info/lettvin/Jerome/WhatTheFrogsEyeTellsTheFrogsBrain.pdf https://web.archive.org/web/20110928024235/http://jerome.let... to Guinea pig eyes https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1564115 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1564115

tibbydudeza 3y ago

From Quora "Each eye has 137 million rods and cones. If we assume 8 bits per pixel and 60 frames per second for 2.5 billion seconds, then the total information from both eyes is 3.3 x 10^20 bits, or about 40 exabytes (40 million terabytes). More realistically, the retina does not transmit pixels directly to the brain. Instead, it detects low level features such as center-surround contrast and movement. Each optic nerve contains about 1 million fibers. We can't see events lasting less than about 0.1 seconds, so let us assume an information rate of 10 bits per second per nerve fiber. This is a total of 10^6 fibers x 2 optic nerves x 10 bits per second x 2.5 x 10^9 seconds = 5 x 10^16 bits, or 6 petabytes. This represents a lossy compression ratio of 6600 to 1. The brain's visual cortex compresses this data further by detecting a hierarchy of increasingly complex shapes such as edges, lines, angles, characters, shapes, and objects such as faces. In a 1986 study in which people were asked to learn and recall thousands of pictures, Tom Landauer estimated that human long term memory is on the order of 10^9 bits, or 1 bit per second. If we assume that 40% of our long term memory is visual (in proportion to the size of the visual cortex), then it suggests that the retina and brain compress visual data at a rate of (3.3 x 10^20 / 4 x 10^8) = 825 billion to 1." https://qr.ae/pKofqQ https://qr.ae/pKofqQ

cbeach 3y ago
Ah so it's the input data over the course of a human lifetime. Thanks!