Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[ChairmanPoo]

I would like to see the source for the claimed number first. Otherwise IŽll file it alongside the "we only use 10% of our braen!" sort of claims.

[Eric Blank]

Bruh I only get 500kb/s max on my phone plan even with a 4g phone and network. The landline internet at the apartment was even worse. Who the fuck gets 11mb/s

[methylatedspirit]

Like it has been said, maybe they meant 11mb of info, information is different from data. Maybe raw data is actually larger.

At any rate 11mb per second mean more than half a giga per minute, and more than 154 gigas per hour, or about 3 and a half terabites of data per day. It does sound like a lot

Let's put it this way: an uncompressed 1920 x 1080 screen at 60 Hz in 24-bit color needs something on the order of 2.99 gigabits a second to run. So it certainly feels really wrong that Britannica says that the visual system "only" gathers 10 megabits of raw data a second.

If that's the compressed data rate, sure, that's reasonable; artificial neural networks can achieve insane compression ratios, so one that's orders of magnitude larger than those currently being used in supercomputers could surely do that. Youtube recommends an upload bitrate of 66 - 85 mbps for 4K HDR video, so it's not a giant leap to assume that our vision (pretending for a second that our eyes see 4K HDR) can be compressed down to 10 mbps by the brain, as amazing as it would be for current video codecs to pull off such a feat.

But if that's the uncompressed, raw data rate, that's where I'd object. At 10 mbps, you can drive an 83 x 83 screen at 60 Hz in 24-bit color. That's not enough for much of anything. Or if you ditch the color entirely, that goes up to 408 x 408 black-and-white. Either way, it's not a good time for anyone.

[ChairmanPoo]

But where are they getting that tidbit from? How was it inferred in the first place?

[methylatedspirit]

But where are they getting that tidbit from? How was it inferred in the first place?

The problem there is that I don't think Britannica cites their sources, nor do they have some kind of 'external links' or 'further reading' section. It's not unusual for encyclopedias to not cite sources, but it becomes a massive pain in the ass to track down where they're wrong if they're wrong, which is what I'm suspecting. For all we know, they're just pulling numbers out of nowhere, and we'd be none the wiser. The whole article is written by one George Markowsky, a Professor of Computer Science, so I guess he might have the answers?

Then again, he doesn't have any contact info on the website, so maybe someone with better Google-fu than me can find the original source?

[ChairmanPoo]

Tbh I have too much on my plate to go down that rabbit hole right now.


... but doesn't it just stink to you? I have the feeling that if we did chase it down we'd find some thing based on some heavily crunched and otherwise unrelated numbers interpretated through a theory that required several unsubstantiated simultaneous assumptions to work

[Lord Shonus]

Doing some digging, I think I found where the original source came from. If you take the number of processors in the eye (1.2 million according to a 1992 paper) and assume they generate 1 bit every 10 milliseconds, you get 120 million bits per second. There's doubt about how much bandwidth the optic nerve can handle, and a reduction to 10 percent of this theoretical capacity gives you the 10-11 megabit claim.

Pretty much everybody I can find who advances this notion immediately deflates it with "the mind doesn't work in bits though, so this is all nonsense because we do not really understand the encoding".

[methylatedspirit]

There's doubt about how much bandwidth the optic nerve can handle, and a reduction to 10 percent of this theoretical capacity gives you the 10-11 megabit claim.

Huh. The optic nerve isn't just a HDMI cable, capable of transporting up to 48 gigabits a second in its latest revision? I think I'm so used to thinking about computers that I'm genuinely shocked that the human brain doesn't deal with such high bitrates.

I wonder if the stuff that transports data between parts of the brain (the "system bus", I suppose?) is also unexpectedly slow compared to the system bus of a CPU. I may be speculating a bit too hard; I've read one too many deep dives on CPU architectures, and I'm making the assumption that stuff in a CPU applies to the brain.

Pretty much everybody I can find who advances this notion immediately deflates it with "the mind doesn't work in bits though, so this is all nonsense because we do not really understand the encoding".

Alright, I think I am speculating way too hard. It's all 'unknown/undefined/not applicable', so anything to come from it must be 'unknown/undefined/not applicable'. I would suppose that the final answer is 'unknown/undefined/not applicable' to that question I posed at the start ("how much data do the senses send into the brain in a second?"), at least for now.

[Reelya]

Like it has been said, maybe they meant 11mb of info, information is different from data. Maybe raw data is actually larger.

At any rate 11mb per second mean more than half a giga per minute, and more than 154 gigas per hour, or about 3 and a half terabites of data per day. It does sound like a lot

Your math doesn't make sense there. If you have half a gig per minute then that's 30 gigs per hour, not 150. You can download BD rips of existing movies that are in that range. I have one HQ rip of a movie that's 25 GB, and this is only 1080p. I'm pretty sure that even that figure of 150 GB per hour is matched by some of the monster 4K rips out there.

Second, the stated amount was 11 megabits per second, not megabytes. The actual calculation is that this equals 1.375 megabytes per second, which works out around 5GB per hour.

[wierd]

I will jump in, when I probably shouldnt.


The optic nerve is not transporting raw optical data.  It is transporting an encoded data stream, and as better educated persons (mentioned in linked papers) have pointed out, we do not fully understand that encoding.

However, artifacts of that encoding can be demonstrated.  The retina actually does a tremendous amount of pre-processing on the visual information it collects, before sending it down the optic nerve to the brain.

https://medicalxpress.com/news/2019-02-early-visual-retina-brain.html


If I were to hazard a guess, the encoding is heavily dependent upon metadata signals generated in the retina, which "cue" the thalamus in how to process the data it has received. 

It would then be less [Raster grid bit pattern] and more like [vector description, size, color values, speed, direction]

This would seem to fit in with some other (much older) research on animal models of object recognition, such as the "Is fly? Y/N" object recognition that occurs in frogs, which appears to happen before the signal leaves the eyeball.

https://www.youtube.com/watch?v=ITY9luBw3_I

[ChairmanPoo]

I don't know if I have the article at hand... but IIRC the information of the optic nerve is not encoded at all. fMRI of the neurons in it allowed for a basic image of what was being seen. I' not talking about the sophisticated machine learning from 5 years ago, I'm talking about a far cruder approach from arond 10-15...

It also made sense. Each neuron in the optic nerve fits one photoreceptor, and thus unless the fibres are crossing over it should be more or less decodeable by checking the activity. It's not a computer cable..

[Egan_BW]

So, there's a disagreement over whether the data gets processed over near the eye, or a couple centimeters over. How thrilling.

[Eric Blank]

Biologists/neuroscientists apparently have lots of arguments that from he outside seem totally trivial. Like whether or not studying whale vaginas is as important as studying whale penises, was an article I read. "Even though there are dozens of research papers on whale penises, there are no records of research on whale vaginas. Their vaginas are interesting too, this one has spiraling skin folds!" Was the premise. The fact of the matter is nobody really wants to get in there and play with whale guts, penises sometimes come to you instead. But who besides marine biologists would argue for digging around in a whale's ass?

Nobody, that's who.

[Reelya]

That kind of thing is engineered outrage. Demanding gender parity when researching extremely non-human species is silly. For example I'm sure there any many more papers researching queen ants and the female worker ants rather than the male drones, and it's nothing to do with researcher bias.

With the whales thing "dozens" of research papers is in fact very little. Busy fields have hundreds or thousands of papers in them, and it's the plain fact like you implied that you can just point a camera at a male whale and have data about their penises. Are we going to send submarine drones up inside whales to get data about their vaginas for the sale of research equality?

There was even a stink about gene drives preventing female animals being born, to control population numbers, the outrage was asking why they don't get rid of the male ones instead. After all it takes a male to have a baby too, so why not turn them female instead? The flaw in that logic should be apparent.

[Egan_BW]

Fund it.