6 ms·

On the other hand, the amount of actual entropic "information" that is processed when you identify a Rubik's cube as such may be nowhere near as much as you thi

On the other hand, the amount of actual entropic "information" that is processed when you identify a Rubik's cube as such may be nowhere near as much as you think it is, and most importantly, 10 bits may be nowhere near as little as you think it is. If we use your example, which is that of identifying an object, we may simply ask the entropy of what the distribution of possible objects-to-be-identified is at t=0, prior to any analysis. Saying we can resolve 10 bits of this entropy per second is equivalent to saying that we can identify one object from a uniform distribution of 1024 per second. Let's suppose this is a low estimate by several orders of magnitude, and that it's really one from a billion objects instead that you can identify per second. Then this would still only be about 30 bits/sec. None of this changes the main thesis of the paper, which is that this is much lower than the 10⁹ bits/sec our sensory systems transmit.

FartyMcFarter 2y ago
But you don't just perceive an object's category (like "cat"). We also perceive a high amount of detail about the object - colour, pattern, behaviour, we make comparisons to past behaviour, predictions of what's likely to happen next and so on. Sure, some parts of the brain don't receive all that detail, but that's necessary for abstraction. If you pumped all the sensory data everywhere, the brain would get overwhelmed for no reason.
- ComplexSystems 2y ago
  That 30 bits was not literally intended to represent only the object's noun category, but even if it did, none of the additional pieces of information you would like to add are going to change this picture much, because what one would think of as a "high amount of detail" is not actually all that high in terms of the logarithmic growth of the entropy. Take color: suppose the average person has 16 baseline colors memorized, and then a few variations of each: each one can be bright or dark, saturated or pastel. That would be about 6 bits for color. If you have an eye for color or you're an artist you may have some additional degrees of freedom. Hell, a computer using RGB can only represent 24 bits worth of color, maximum. I am going to suggest this stuff gets cognized less than 10 bits worth for the average person; let's just say 10. Now, of course, people can memorize more than one color. If colors are independently distributed uniformly at random, then processing N colors requires 10N bits. But of course they aren't, so the entropy is less. But again, let's just say they were. So how many color combinations can you process per second? I would say it's a bit of a challenge to memorize a set of 10 arbitrary drawn colors shown for a second. Most people couldn't continuously do that at a rate of 10 colors per second. That would be 100 bits/sec of info. The point is that you really don't perceive all that much. You show the average person a Rubik's cube, there is no way they're going to remember the exact pattern of colors that they saw, unless the cube were solved or something. They will perceive it as "multicolored" and that's about it. Adding behavior, texture, etc doesn't change this picture. None of this stuff is even close to 10^9 bits of entropy, which would be 2^1,000,000,000 different equally likely possibilities.

FartyMcFarter 2y ago

But you don't just perceive an object's category (like "cat"). We also perceive a high amount of detail about the object - colour, pattern, behaviour, we make comparisons to past behaviour, predictions of what's likely to happen next and so on. Sure, some parts of the brain don't receive all that detail, but that's necessary for abstraction. If you pumped all the sensory data everywhere, the brain would get overwhelmed for no reason.

ComplexSystems 2y ago
That 30 bits was not literally intended to represent only the object's noun category, but even if it did, none of the additional pieces of information you would like to add are going to change this picture much, because what one would think of as a "high amount of detail" is not actually all that high in terms of the logarithmic growth of the entropy. Take color: suppose the average person has 16 baseline colors memorized, and then a few variations of each: each one can be bright or dark, saturated or pastel. That would be about 6 bits for color. If you have an eye for color or you're an artist you may have some additional degrees of freedom. Hell, a computer using RGB can only represent 24 bits worth of color, maximum. I am going to suggest this stuff gets cognized less than 10 bits worth for the average person; let's just say 10. Now, of course, people can memorize more than one color. If colors are independently distributed uniformly at random, then processing N colors requires 10N bits. But of course they aren't, so the entropy is less. But again, let's just say they were. So how many color combinations can you process per second? I would say it's a bit of a challenge to memorize a set of 10 arbitrary drawn colors shown for a second. Most people couldn't continuously do that at a rate of 10 colors per second. That would be 100 bits/sec of info. The point is that you really don't perceive all that much. You show the average person a Rubik's cube, there is no way they're going to remember the exact pattern of colors that they saw, unless the cube were solved or something. They will perceive it as "multicolored" and that's about it. Adding behavior, texture, etc doesn't change this picture. None of this stuff is even close to 10^9 bits of entropy, which would be 2^1,000,000,000 different equally likely possibilities.