I've seen it repeatedly stated that the eyes process 36KB of information per hour. I haven't been able to track down the source or justification of this number though.

I think this is a fascinatingly deep question. It is a surprisingly difficult quantity to estimate because you need to understand the coding scheme (i.e. how will each spike from each fiber will be interpreted at the receiving end) to truly understand the bit rate being transmitted.

If you want to come up with a maximal limit for the human eye, you could assume that each spike carries 1 bit of information per 10 ms time window, so that each fiber could contribute 100 bits/second. Multiply by the number of fibers in the optic nerve (about 1.2 million, source: http://www.iovs.org/content/33/6 look for Jonas et al 1992 ) and you come up with 120 million bits per second as a theoretical maximum throughput per eyeball.

I suspect that the optic nerve is probably physiologically incapable of supporting that rate of transmission however, so I would suggest you stick with Geoffrey's far more conservative answer. And in any case, without an understanding of the encoding scheme used by the visual system, the number is almost meaningless.

de Ruyter van Steveninck and Laughlin reported the information transmission of blowfly photoreceptors at as much as 1,650 bits per second in their paper "The rate of information transfer at graded potential synapses" in Nature vol 379, p. 642, 1996.

http://www.princeton.edu/~wbialek/our_papers/ruyter+laughlin_96a.pdf

One can recognize about 4 complex objects (like houses) in a single glance,

but many more than 4 simple objects (say, poles) in the same length of time. So there is no fixed answer, in terms of number of items. Therefore one needs

to use a theoretical approach, such as entropy or information (e.g., bits/second).

But the difficulty in estimating 'throughput' from a bits/sec point of view is that one must estimate the entropy of both source and receiver to figure the throughput..

No-one knows how to specify the number of bits in the typical visual scene (the 'source'), let alone how many bits there are in the receiver - the brain and its sensory representation of the scene. One can only get a number for a

constrained situation, like reading random letters from a fixed alphabet (of 26 letters, say). So this remains a 'hard problem' and despite decades of work, it has been laid aside as unsolvable based on current knowledge, even though one expects it eventually will be solved. (I'll be interested to know if others agree with this general picture of things.)

Only a small amount of the sensory information our modalities collect becomes conscious, because certain neural systems report to our awareness whereas other do not - removing the cerebellum for examples does not seem to impair consciousness. However, does consciousness add to the total throughput? Awareness certainly influences what and how much information we attend to in the environment.

Tononi has proposed an information integration theory of consciousness which states: the quantity of consciousness corresponds to the amount of integrated information generated by a complex of elements, and that the quality corresponds to the informational relationships within that complex.

So why does some information become integrated and some not? Maybe attention increases the sensory bit rate for a certain modality, while decreasing for others or even within the same modality. Think of the cocktail party problem - you can choose which voices to attend to and they become clarified over the other voices. So is your sensory system generating different bit rates depending on what you're paying attention to? Or is the sensory bit rate constant just directed by attention? Or does this happen after the information enters the neural system? The Raichle quote above suggests not because once the information is in the brain it quickly degrades - so the brain (you) must choose what sensory information to get from the environment. But the selection process of what to attend to in the environment seems to happen without conscious awareness (see Carasco's or Llinas's work for example).

Then you have the attentional blink paradigm which seems to indicate that there is a refractory period for at least perceiving new information. Does this information get sensed but not reported further up to awareness? Some experiments suggest that this is so and that our senses take in much more information than we're aware of (see Dehaene).

I think the relationship between what is sensed and what is perceived is actually a very complex bottom-up and top-down process. Is there some maximal bit/rate of the sensory system? Or is it even constant? Maybe it's highly context dependent.

Meta-Comment: My approach, like Smarts', is to look only at the gross function. Others look only at the information rate within the channel (Plass, Wyble). But surely the two must be bought together. For example, the resolution of the photoreceptors is probably much finer than the detail needed to encode the environment for all normal visually-guided behaviors, so the theoretical 120 m bits/sec is meaningless, by itself. The system can (and actually does) average over numerous photoreceptors at the ganglion cell level, and this probably doesn't doesn't reduce information transmitted (it may even increase it a bit, by reducing noise). To put this on a scientific basis, though, one needs to know what is in the environment, as well as what is transmitted by the optic nerve. As I said, since we don't know what the environment offers, we cannot do our sums; yes, we can now estimate roughly how much averaging occurs (by looking at ganglion cell receptive fields), but we don't know how much the system ought to average so as to improve S/N by reducing the noise level, N, without reducing signal power, S, by too much. (I think this same argument applies to all sensory systems.)

I concur completely with Adam here. I would hate to go on record as saying that we "see" at 120mbits/second. I was just illustrating how one might come up with a number using the known properties neurophysiology.

But that number is quite meaningless without a deep appreciation of the visual system's data compression algorithm in the retina, and its processing algorithms on the other end of the optic nerve.

This tutorial by Dave Touretzky might help to provide a clearer picture of the importance of information theory in answering this nearly unanswerable question:

http://www.cs.cmu.edu/~dst/Tutorials/Info-Theory/

Here are two papers on how the information capacity is related to axon caliber, also from the visual system:

Why do axons differ in caliber?

Perge JA, Niven JE, Mugnaini E, Balasubramanian V, Sterling P.

J Neurosci. 2012 Jan 11;32(2):626-38.

How the optic nerve allocates space, energy capacity, and information.

Perge JA, Koch K, Miller R, Sterling P, Balasubramanian V.

J Neurosci. 2009 Jun 17;29(24):7917-28.

To add to Reeves' point about gross function and channel capacity: think of an illiterate person looking at words printed on a page versus you reading. How many bits are transmitted to the illiterate person compared the number of bits transmitted to you? Presumably - (but maybe this has been looked at empirically) - at the level of photoreceptors, optic nerves and axons, ie at the sensory level, literacy would not make a difference. But in addition to the sensory bits being transmitted, the literate person also receives the bits encoded in the language that is written. We know the brain recycles neuronal structures when we learn to read - does this extend to the sensory level?

Hi Helgi,

It's a bit tangential, but in case it helps, - in investigating dyslexia issues they use a measurement of brain processing across sensory modalities such as the P3b amplitude, e.g. this article, in

Brain research: B R A I N R E S E A R C H 1 3 8 9 ( 2 0 1 1 ) 9 3 – 1 0 2

P300 as a measure of processing capacity in auditory and visual domains in specific language impairment Julia L. Evansa,⁎, Craig Selingerc, Seth D. Pollakb,c

a School of Speech Language Hearing Sciences, San Diego State University, San Diego, CA, USA

Helgi,

Very interesting question. As others stated, the most difficult part is that we do not know exactly the way information coded. Though neurons are sending spikes through fibers, it may carry far more stuff than 1 bit / spike. As bits can be carried with a single spike, the timing of the spike, the density of the spike, the interval between two spikes, and corrlation of spikes among many fibers. The combination could be a lot more complex then we already know. And we still don't know if information was carried in bits as if in computers. The brain may well have other ways of organization. Probably, this can be estimated by looking at the changes in the environment and how much we have perceived. However, that is again only an estimation.

In addition to what Balázs Hangya wrote (Balasubramanian et al.):

the neuroscientists estimate that the human retina can transmit

10 million bits per seconds (to the brain, I suppose).

http://www.eurekalert.org/pub_releases/2006-07/uops-prc072606.php

A thought experiment. How long does it take to store 10 000 bits

into long term memory? Attached is a matrix of 25*80 random

letters. Each letter needs about 5 bits to be stored (2^5 = 32).

I estimate any person needs several hours if not days to learn

the page by heart, so the information transfer rate to human

long term memory must be in the order of 1bit/s.

As far as I know Ebbinghaus did such experiments in reality

in the 19th century.

Regards,

Joachim

I did not take the original question to be asking specifics of neural systems per se, but to be asking the bigger question such as: How much of the Electro-Magnetic spectrum out in the environment can a human perceive, in their entirety of sensory modalities?

Is that more of the question?

I have been asking this question for many years. I believe our construct of perceptible abilities is what gave us the ability to use the bigger brains some mutations had afforded some, and helped us differentiate ourselves from the other primates.

If we look at our senses, we see that as humans, we have actually culled or attenuated some senses (smell for instance) while amplifying others (namely vision.) Interestingly, this attenuation has not abolished the nose, or turbinates, as they provide other advantages in breathing that maintain their utility, but, the brain has adapted to human's ability to use its visual system better.

Even though others here have posed formidable data reduction models, this really does not account for how the mind can build a picture from sensing the smallest amount of information, and therefore convey the same representation from divergent data streams. Just look at the night sky's constellations. Here we see the faintest of points of light create mental images and stories that do not rely on using anything but the very basics of the actual carrying capacity of the visual system. What is the bit depth of the big dipper?

I feel the human senses, our windows on the world, in combination, are the greatest swath of electro-mechanical environment gathering tools present in the animal kingdom and this is what gave us our big brains and eventually our mind.

Even though some animals have different perception capabilities (snakes heat sensing,) those abilities were not enough when interacting with their other senses (which are low capacity such as touch) does not afford it a true bigger window on the world, ergo, snakes remained a small brained animal. Those mutations that offered a bigger brain did not have the stuff (sensations) to feed it, so a bigger brain provided no clear competitive advantage. The ability to sense goes hand in hand with the ability to think.

To the extent that Helgi is interested in the amount of information that passes through a sensory system, a good way to frame the issue is to consider a case of rapid serial visual presentation (RSVP), in which a series of stimuli are shown in rapid succession.

For example, if I were to show you a series of 1024x768 natural images (about 8 million bits each) at a rate of 1 image per 150ms, and ask you to spot the picture of the dog, you could easily detect and remember the dog, but your memory of the other images would be relatively poor (Potter 1976 Short-term conceptual memory for pictures, JEP:HPP).

The actual rate of processing is not indexed by the image you remembered, but by all of the images that you discarded. In order to detect the dog with nearly perfect accuracy (and having no false alarms as well), then it must be the case that your visual system decoded all of the non-dog images as well, at least to the level at which they could be said to contain a dog or not.

Therefore, one might say that your processing ability is at least 1 full color image per 150ms (or about 50 million bits/sec!) But the answer is not as simple as this.

First, your familiarity with those images is essential to allowing you to process them so quickly. You have developed highly specialized neurons that can decode real-world images quickly.

Second, most of those pixels form textures and contours which your visual system collapses into simpler representations; so that total amount of information present in each image is substantially less than 8 million bits.

To illustrate, if I ask you to spot the attached image (10,000 bits of information), among a series of similar patterns, your accuracy would probably be very low indeed, even at a rate of 300ms per stimulus.

After a few years of training, you might improve considerably but you would probably never rise to the same level of accuracy as for natural images. (this experiment has not been done to my knowledge, but someone please correct me if I'm wrong).

The beauty of the senses are that as they progress from organ to brain, each successive step transforms the information from binary to analog ever so slightly, such that by the time it reaches the cortex and is primed to be further processed for memory consolidation and emotional content attachment, it is a stream of experiential gestalt, instead of a bunch of blips. So the measuring that is going on in bits/sec is for naught. We synthesize the data streams simultaneously, which in effect yields a new perception in its own right. But, interestingly we do not have words to convey these streams of perception. How would you describe being in a flight simulator? Only by explaining how each of the component perceptions felt. But, the experience is more than the sum of its parts.

The blotchy image presented above for visual inspection is useless as a true stimulus for the task at hand. The seemingly randomness holds no more information than if white and black were segregated into two big sections. If the underlying detail is of interest, and differentiation is required, then it is akin to looking for waldo. Fine acuity is brought to bear now extracting detail that before was not asked of the original "sensation" of the image. If we were tasked with, lets say, recognizing a robot face in the dots, we would need to intensify our focus of the imaging such that we make the relativity of the black and white areas to each other important, not just the impression of a splotch of black and white dots. Perception, once again, is different from sensation and helps to control the act of sensing.

So since the perceptual intent controls the sensation focus, the original question: "...how much information (raw data) is generated (sensed) by the sensory system (or the major modalities) prior to perceptual activities" needs further control and clarification.

re: RSVP- Wyble: jpeg compression shows that a much reduced

image looks just like the fully-fledged 8 million bit image. In RSVP, one probably processes even less of each image, so the compression could be even more before one noticed. Specifying the 'bits' in the environment is possible, but not simple.

I estimate that there are reversible the reading process and the writing process. The difference could be significant.

Sorry: I estimate that there areNOT reversible the reading process and the writing process. The difference could be significant.

Thanks for all the answers, there are some very useful references in there already. To clarify my question and motivation further, I am working on attention mechanisms for AGI (artificial general intelligence) systems. If we take a very functional view of a human being, I am interested in the size of the "data stream" that enters the nervous system through all the various sensors before any kind of perceptual processing or information reduction. So my question may be more related to the complexity of the real world as perceived by us. Ultimately, human-like AI systems (e.g. humanoid robots with general cognitive abilities) will have to perform (and learn to perform) tasks in the real world and deal with a similar amount of sensory information as humans. This is why it seems interesting to me seek out or create quantitative estimates of this environmental complexity.

Another interesting approach has been proposed by Karl Friston. His free-energy principle is toward a unified theory of brain function. Basically the organism is at all times trying to reduce the 'surprise' of both its internal sensory states and its environment to maintain homeostasis. So he says for example that a 'fish out of water' would be surprising both emotionally and mathematically. But what is surprising for one animal is not for another because each species is adapted to certain types of environments. In other words, Friston points out, that biological organisms must violate the second law of thermodynamics by keeping their long term average entropy low. So all you have to do in AI is design a robot that can avoid being surprised as often as possible ;)

As far as quantifying the data stream that impinges on our sensory system - I think many of the responses posted, my own included, have tried to argue that you might have the cart before the horse so to speak. If Friston's ideas are in the ballpark - then basically the organism is designed to minimize the amount of data required to stay in a low entropy state. So perceptual processes guide the nervous system toward the most useful and light weight data - because it is (and has been for the history of the evolution of that species) on average the least surprising.

http://www.fil.ion.ucl.ac.uk/~karl/The%20free-energy%20principle%20A%20unified%20brain%20theory.pdf

Helgi,

I am confused. Why would you build AI robots to match human sensory capability only?

"Ultimately, human-like AI systems (e.g. humanoid robots with general cognitive abilities) will have to perform (and learn to perform) tasks in the real world and deal with a similar amount of sensory information as humans."

I would think that a new system would not have to be saddled with the biochemical evolutionary parsing constraints that has occurred in humans' ability to perceive the environment. I would think one would make an AI system to have all manner of sensing capabilities, and then to "perceive" like a human, the system would, in actuality, turn off or attenuate some of the sensing systems that are not of, or outside human capabilities.

Would you not afford your AI system of IR, and echolocation, and much better olfactory sensing? Why would you only strive to put in human sensing traits/capabilities? Put them all in, and turn on and off systems/capabilities to mimic the qualia of other organisms.

Claude, I completely agree with you and have no interest in biologically inspired sensory limitations with regards to AI systems. However, it is useful for my work on attention for AI to be able to argue for its necessity by citing the example of how much information is sensed by humans. We already know something about human working memory, so it would not be hard to get some idea of the information reduction that occurs in humans, which attention is largely responsible for.

i think that's a good point Dr Albertino - if your robot had an IR camera and some kind of enhanced vision system it could find its way in the dark, in the fog etc. There are huge parts of the electromagnetic spectrum humans can't perceive but that current technologies can. of course what we want the robot to do is "show" humans these invisible areas of the spectrum. or perhaps eventually merge enhanced vision systems with human biology - put some kind of nano-engineered IR cameras in our own retinas and implant chips and processors to directly translate the images to our own visual system. then we could turn on IR vision for night driving or walking. likewise with echolocation. so how many bits would those additional streams of sensory input add?

sorry Albertario.

Just a note, I am not a Dr., but I do play one on TV. ;)

I still think the whole gestalt of "sensing" is too abstract for simple "data throughput" measures. Information reduction is not the question if we do not really understand what is being sensed? the information may burried within subtle nuances of simple 1 bit black or white lines, but that does not mean the information is only 1 bit. I still see the question as trying to measure wetness with a ruler.

Perhaps for similar reasons as Claude here, I worry about this question because it seems to come from a perspective that biological systems are the kinds that submit singularly and uniquely to any given measurement frame. In order to index "throughput" in these engineering-like terms, you need to reduce the system you're analyzing to a single dimension. "How much?" has to become "How many [units on some dimension]?" And what Claude has ably been suggesting is that, try as we might to probe the system objectively and plainly, biological systems (or just humans, to keep it simple and to the point at hand) tend to thwart all these probes with its many nonlinear transforms. Biological systems are great at reshuffling structure from one measurement space to another--and sort of scattering our accounting efforts to the winds.

I think the jury is out on enough about memory, let alone working memory, to discourage this sort of inquiry. Information for sensing is not reducible to Shannon formalisms--as James Gibson (I think, another Cornellian, yes?) was wise to point out.

But it is biological systems (humans) that invented the measurement frames - i forget who said it but it's that paradox of "if our brains were simple enough for us to understand we would be too simple to understand our brains."

No one is saying that we cannot understand it EVER, just with the linear engineering principles it seems to be something that is outside the bounds of these very limited descriptors.

One reason is that the sensation is only a "sensation" prior to its recognition, at which time it becomes a perception with all the myriad of emotional and memory attachments to it.

So, if the question was parsed to, what is the bit decrease or increase in imagery content as a visual image gets projected on the retina, compared to its neurological representation at each of the way points along the optic pathway to the cortex,then we might be able to use the linear tools to measure these things. But, to ask the general question "how much information (raw data) is generated (sensed) by the sensory system (or the major modalities) prior to perceptual activities per some unit of time." leaves it up to the wind to decide which way point to measure up to.

There are also sensations we do not recognize as perceptions, such as the amount of light at the Super-chiasmatic nucleus that governs our temperature rhythm. Would that be "information" gathered by the eyes, or is that not valid information? Should something be subtracted from the amount of raw data at the eyes since some of it goes to this non-percepting endpoint? Or is a change in temperature a "perception?"

Reductionism only works when there is proper calibration of the two systems being compared. There is no way to "calibrate" a sense against anything.

I think that we don't even understand measurement frames. Whatever the general case of "measurement" is, our crude version is only the one we're able to wrap our minds around. I don't think it's a paradox. No measurement system is perfect. By definition it has constraints, otherwise it doesn't work. The stuff we do that we call "measurement" may only be our folk-tale version of what millions of other systems (biological, physical, or whatever other kind) do in adapting and coordinating with their contexts. But we don't even have to go that crazy-sounding extreme to see that the human system regularly flaunts its own attempts at measuring itself. And I would say that any system coordinating with its context will flaunt measurement in just the same way. I don't advocate not doing measurements of course. I just hope we keep realistic expectations as to how much and what can be learned from each one.

Claude, if I could vote up your comments more than once each, I would!

The performance limits of the human eye itself is estimated to be somewhere around 6 Megabits per second, taking into consideration the actual measured resolution and speed of response of the eye. In The Oxford Companion to the Mind ISBN: 0198662246, Robert Ditchburn describes the reasoning as follows:

“Accurate data is available only for the central region of the retina (a circle whose diameter is 2 degrees in the visual field) and the total information capacity of the visual system is about ten times that of this central circle, the fovea. The number of points that can be resolved within this central circle is about 10^4 (i.e. 10,000 pixels). When the eye is working near the limit of resolution, contrast discrimination is poor and at the limit of resolution, a person can distinguish black from white but not discriminate any intermediate shades of grey, therefore he associated just one bit of information with each of the resolved points. For the central region the highest rate that flicker can be observed is about 50 cycles per second, so that the information capacity is 50 × 10^4 = 500 kbits per second”. Measurements of contrast sensitivity with coarser sinusoidal patterns with various shades of grey gave almost identical results.“Perception of structure in flashes and in afterimages”, (Ditchburn, R. W., and Drysdale, A. E., Volume 13, Issue 12, December 1973, Pages 2423–2433,“Information capacity of the human eye”, Jacobson, H., Science 113, 292–293 (1951)).

Neither of these estimates includes colour information, but it is estimated that the additional information due to colour is no more than about 20 per cent, making 600 kbits per second in all for the central region. Given the estimate that the total information capacity of the visual system is about ten times that of this central circle, the fovea, this gives a total figure (per eye of around) 6 Mbits per second. The raw capacity of the ear is considerably less. The rate that we can learn anything new through our senses is much less, ~20 bits per second.

And if anyone is still interested, here is an article on the limited information rate at which we are connected to the world through our senses: What do we really know about our human interface?

and see www.humanbottleneck.com

www.humanbottleneck.com

https://medium.com/@richardepworth/what-do-we-really-know-about-our-human-interface-2fa808c130a0#.399q6p7x2

In my opinion, calculating information transmission from assumptions about bits per second in the retina or optic nerve is fraught with error.  Here are some of the igored factors:  (a) it is not just the mean spike rate, but the temporal distribution of spikes, that carries information in the optic nerve; (b) nearly all environmentally critical information can be carried by a  visual system with much patchy retinal loss (as opposed to  areal loss concentrated in the fovea, say, like AMD, or night vision, which is devastating for tasks that require high acuity), because adjacent photoreceptors are highly redundant,  (c) the calculations typically ignore critical visual properties such as adaptation, spatial frequency encoding,  and noise suppression, and (d) contrast encoding is vital, not encoding the absolute light level, which is critical to the organism but requires very few fibres to specify it.  Estimates of information :"per pixel" which I have seen don't seem to realize that it is information across groups of receptors, not within individual receptors, that matters. I think if the retinal surface was continuous, rather than discrete (it has to be discrete, because of the waveguide properties of the rods and cones, but one can imagine this physical constraint did not exist) the entire 'pixelization' analogy would never have been dreampt up.

Adam Reeves. 

Adam, the Bit-rate approach is most valuable when estimating the maximum sensory capacity for novel information, using a “Black-Box” approach, i.e. measure information in, confirmed via information out. This sidesteps all the details of the internal structure. However the following might be of interest.

The human eye is estimated to interconnect with our brain at somewhere between 6 and 10 Mbit/s, both from measurements of its capabilities and neurologically. This figure represents the maximum biological connectivity with our human brain, (and limits our ability to recognise what we expect through our senses).

It is estimated to be somewhere around 6 Megabits per second from the measured resolution and speed of response of the eye. Robert Ditchburn [1] describes the reasoning as follows:

“Accurate data is available only for the central region of the retina (a circle whose diameter is 2 degrees in the visual field) and the total information capacity of the visual system is about ten times that of this central circle, the fovea. The number of points that can be resolved within this central circle is about 104 (i.e. 10,000 pixels). When the eye is working near the limit of resolution, contrast discrimination is poor and at the limit of resolution, a person can distinguish black from white but not discriminate any intermediate shades of grey, therefore he associated just one bit of information with each of the resolved points. For the central region the highest rate that flicker can be observed is about 50 cycles per second, so that the information capacity is 50 × 104 = 500 kilobits per second”.

Measurements of contrast sensitivity with coarser sinusoidal patterns with various shades of grey gave almost identical results [2]. Neither of these estimates includes colour information, but it is estimated that the additional information due to colour is no more than about 20 per cent, making 600 kilobits per second in all for the central region. Given the estimate that the total information capacity of the visual system is about ten times that of this central circle, the fovea, this gives a total figure (per eye of around) 6 Mbits per second.

Remarkably similar neurological figure of ~ 10 Mbits per second is obtained from the study of ganglion cells carrying information from the retina [3].

1/. The Oxford Companion to the Mind”, (1987), Edited by Richard L. Gregory. ISBN: 0198662246, (2004)

2/. “Perception of structure in flashes and in afterimages”, Ditchburn, R. W., and Drysdale, A. E., Volume 13, Issue 12, December 1973, Pages 2423–2433 “Information capacity of the human eye”, Jacobson, H., Science 113, 292–293 (1951)

3/. K.Koch et.al. "How Much the Eye Tells the Brain," Current Biology 16 , 1428-1434, (2006).

Www.humanbottleneck.com

Richard

Www.humanbottleneck.com

"Information" is a cognitive function. Seeing red light or green light is basically the same sensation, yet the up stream information they each hold can be totally different.

Also, we sense in parallel not serially, so bitrate is poor metric.

CC: Jon Plass answer: It's Marcus Raichle (not Reichle)

"Two Views of Brain Function", Trends in Cognitive Sciences

DOI: 10.1016/j.tics.2010.01.008

thanks for the calculation of bit rate in vision. An important next question concerns receptive field (RF) overlap. If RFs overlap as much as they combine cone responses, then every cone gets one cell in the brain, i.e., 1:1; but if the overlap is less than the typical RF size, then information is combined and the bit rate goes down. I don't know what the answer is to this. I suspect that in the fovea it is 1:1 , but less efficient in the periphery.

Do we have an agreement that these data estimates strongly suggest that the visual cortex receives an impoverished representation of the world? Yes, we need better estimates for both conscious and nonconscious processing rates. I would welcome anyone's data from any effort to quantify these systems. thanks

I have searced everywhere (or hopefully not) for a relevant figure of the interoceptive input to the brain in bits per second. Mainly through vagus that has 500 million receptors connected to brain.