Recently I decided to learn more about brain structure and organization, especially in humans. As modularity is a key concept in complex systems, a key question is: what organizing principles explain which parts are connected how strongly to which other parts? (Which in brains is closely related to which parts are physically close to which other parts.) Here are some things I’ve learned, most of which are well known, but one of which might be new.
Sorry, it appears my posts are appearing in reverse order. Please read the third post below first. Then the second, and this post last.
Organizing principle #3: the right brain processes bottom-up, the left top-down.
I am in total conflict with you here. Both sides of the brain use both bottom-up *and* top-down processing. The function of the nervous system is to allow the organism to respond to the environment. In the bottom-up/top-down paradigm, sensory input is bottom-up processed as signals travel to and within the brain, then the brain top-down processes the signals to send back to muscles to cause behavior. As mentioned in #2, the sensory signals cross to the opposite side of the brain for the more concrete levels of processing. So input from the right side of the body is processed in the left brain. The left brain then sends signals to the right side of the body for movement, and vice versa.
The closest version of bottom-up/top-down is the division of the brain into sensory (bottom-up) areas in the parietal lobes (above and behind the ear) and motor (top-down) areas in the frontal lobes (above and between the eyes and ears).
One has to be careful about bottom-up/top-down conceptualization, though because it implies a locus of initiation. The nervous system is constantly processing in both directions. If there is a starting point, it is the sensory input from the environment.
As to left-right differences, as I discussed in my original post, the major difference appears to be language processing in the left. But these areas are highly connected with areas on the right (like server farms on different continents).
Organizing principle #2: collect things at similar levels of abstraction.
#2A: the rear of the brain is concrete, the front is abstract.
Yes to #2 (subject to discussion of ‘near’ in #1, and if you are careful about the meaning of ‘abstraction’.) Information in processed significantly even in areas outside the brain. For example, the retina of the eye has already identified the edges of objects before the signal leaves the eye to travel to the brain.
Mostly no to #2A. The fact that the rearmost part of the brain, the occipital area, is more ‘concrete’ is an accident of the placement of the special senses of vision. For some reason the nervous system tends to place association (abstraction) areas on the opposite side of the body from the sensory input they process. Touch on the right side of the body is processed by the left brain. Visual input from the front of the body goes to back of the brain.
However, the *dominant* organization of ‘concrete’ and ‘abstract’ is the center of the cortex (a strip across the top from ear to ear) being the most concrete. As you move away from that strip, both forwards and backwards, the brain gets more abstract.
But just because they are far apart doesn’t mean the abstract areas in the front of the brain are less connected to the abstract areas in the back. An analogy is that the server farms of Google are spread across the world, but the connections among the server farms are arguably stronger than the connection of a particular server farm to a house a mile away.
I will address your hypotheses more directly. On my reading you have three main hypotheses, or organizing principles, related to brain function. I will address each in a separate response.
Organizing principle #1: related functions are more connected to each other.
Yes, but more connected doesn’t necessarily mean *near* each other. One general principle of the nervous system is that there are at least 3 different types of connection. The most ‘dense’ and slowest is where the axon (the output) of the neurons connect to many nearby neurons. Signals spread in a wavelike fashion.
A second connection is where signals are collected from several thousand neurons and sent to other, relatively nearby units. The cortex of the brain is formed of 6 layers of sandwiched neurons, some layers are more ‘local’ in function and often form a column of related neurons. Others layers communicate with other columns.
A clumsy, but serviceable analogy, is that each neuron is a microprocessor, the locally-related neurons in a column are a server, and an area of the cortex encompassing several thousand columns is a server farm.
Finally, the third connection type is fast connections among ‘server farms’, that can span the entire brain. The connections of the corpus callosum, which crosses from left to right brain and back, are of these third type.
Disclaimer: say no to ad hominems. Oh, no, I'm not saying that you, Robin Hanson, are unaware of details. That was poor phrasing by me, sorry. I tried to show that human brain (mostly young areas, such as the frontal lobes developed in historical time of behavioral modernity) as an information processing entity can recognize/formalize only tiny fraction of what's going on behind the gates of the ACC. And this tiny fraction of accessible information that we, humans, can study is an end result of several interrelated neurological ensembles. There is evident separation, of course. An much more than just binary separation. So, our conscious self employs young and articulate frontal cortex to interpret those signals from our ancient parts of the brain. That's the point. Better now.--In no way I was trying to overwhelm you with details (I suck at writing and don't really know how to explain brain anatomy without mentioning certain things). Nor did I try to prove something here, on your blog. Just wanted to say hi, question some alternatives that might be complementary to the binary concept. I have no further comment.
I trust you're aware of the details, but whether your theory successfully explains them can't be evaluated by this post. [Not to carp, but some such disclaimer might have been useful.]
[The view that the left brain is deductive and the right inductive is fairly common. But that this represents the basic principle of its organization apparently novel.]
While Jo Mo doesn't state his objections crisply, some can be inferred. One such objection (which I thought of too) is that the capacity of the brain to reorganize itself may be in tension with the conclusion that the design principle is based on keeping things separate. (A testable inference might be that when the brain reorganizes functions from one side to the other the organism will experience more self-sustaining credit cycles than one with a brain not reorganized. An unexpected consequence of minor brain damage.)
I of the Vortex by Rodolfo Llinas (2002) says as much as can be usefully said about the core function of the brain. From his premise we can begin to understand why our large brains can be subject to useless fantasizing such as speculations without utility.
On what basis do you claim that I'm unaware of details? That I didn't give a long list of details like you did? You think blog posts and their comments should be contests to show who can list more details?
That's the point. You can contemplate on general patterns (top-down) while simultaneously being unaware of a lot of gritty details underneath those patterns (bottom-up). A very smart design that reduces a lot of noise.
Data: Meditation increases connections. Assumption of mine: Advanced meditators display symptoms of what you call self-supporting cycles. Plausible cause of self-supporting cycles: increased excitatory connections, not inhibitory
The claim that a big function of the connections between hemisphere is to prevent self-supporting cycles is consistent with that total connection size depending on many factors.
"The main risk of mixing credit directions is creating self-supporting credit cycles not well connected to real needs. This may be why the connections between the two hemispheres are mostly inhibitory, reducing activity." If that were true, what would it suggest about increased connectivity between hemispheres in meditators? Assuming meditation involves learning to perceive sound as sound (not windy leaves of a tree, not bus, etc.)? Seems to suggest that excitatory connections are being formed, or perhaps your hypothesis tends towards being wrong Related: https://www.ncbi.nlm.nih.go...https://www.ncbi.nlm.nih.go...
I appreciate your writings. If you are going to further pursue brain structure and function, I hope you will be careful about your sources. There is a lot of magical thinking about brain function, especially the right/left differences.
I have done research in neurophysiology, and part of my function when I was at NIH was to keep up with the neurobiology research. Most of my experience, however, has been more clinically oriented with patients with neurological dysfunctions.
If you look at the big picture of neurological function, the system works to receive sensory input, process that input, and create functional (mostly motor) output that enhances the organisms response to the environment. If you stimulate one side of a worm, it's nervous system will contract muscles that withdraw that side away from the noxious stimulus. The functions are preserved in humans, and generally operate at the spinal cord level.
As nervous systems evolved and became more complex, new functions have been layered on top of the old ones (I hesitate to use the term pathways, because they lead people to conceptualize wires, or the old style phone systems; the functions are more comparable to distributed communication like in the Internet.) The organisms reactions to sensory input have become much more complex, and separated from the input in both time and space. A high level of separation has often been labeled as abstraction.
There is a high level of processing throughout the nervous system. Humans have 100+ billion neurons, each neuron functions as a processor with several hundred to several thousand inputs (from dendritic connections to upstream neurons) creating a continuously variable output (axonal firing rate) that signals downstream neurons.
At the risk of error through over-simplification, the function of the spinal cord is the first few levels of processing of sensory input from the body and creating simple movements (a cat's hind limbs can make stereotypical walking movements even with a totally isolated spinal cord). This already highly processed information passes to the brain stem, which adds in specialized senses such as taste and hearing, and basic visual input. The brain stem output is mostly homeostatic functions of the body (blood pressure, breathing, sleep, etc.).
The midbrain adds in more complex 'rehearsed' movements, e.g., a batter hitting a baseball. Some types of memory (e.g. procedural) exist at the midbrain level (and if you have a broad definition of memory, even lower in the nervous system). The cortex is the most 'abstract', and appears to have evolved to add flexibility and improved pattern recognition across time. A fairly accessible example of the functions of the cortex is to compare cats and dogs. Cats have smaller a simpler cortex. While they are superb at highly stereotypical behaviors, the system can often be 'fooled' again and again, e.g., a cat will pursue a laser pointer dot, as it triggers the 'prey-pursuit' functions. The more complex cortex in a dog allows more specific training, more 'sociability', etc.
The human cortex has two main input areas which are somatotopically arranged: the visual system (in the back of the brain, the occiput), and sense of touch/position (a strip arching across the top of the brain from ear to ear). These sensory inputs then spread into sensory association areas, which, as you posit, become more 'abstract' the farther you get from the primary sensory areas. This spread travels 'forward' from the visual areas, but 'backward' from the somatosensory (touch) areas. These 'parietal' areas includes the back half of the brain. The areas 'furthest' from the primary sensory areas are the temporal lobes, which extend laterally and forward from the parietal lobes. The temporal lobes contain the most abstract functions such as language.
Up to this point, most of the brain functions are the same on both sides. The major difference in right and left brains is in the functions of the temporal lobes, with the left side usually containing most verbal language function, and the right containing non-verbal language and non-language communication functions. If you look at the detail of studies on right/left brain differences, the ones ascribed to left dominance are those with tests that require verbal language (using/remembering words). Although we thing of communication as being verbal, over 90% of human communication is non-word based, and over 60% is non-verbal. For example, think of the difference in texting versus a phone call. Texting loses all the non-word communication such as voice tone, speed, pitch, etc. Now look at the increase in communication that comes from face-to-face interaction, where non-verbal information such as facial expressions, posture, etc. are added. Damage to the right brain causes loss of skills in those non-verbal domains.
The output from all of these sensory association areas (including the temporal lobes) transits forward to the motor association areas of the frontal lobes, which formulate behaviors at various levels of complexity. The most 'abstract' motor areas are in the front, and the information flow moves 'backward' to the primary motor areas, which are just forward of the primary somatosensory area and similarly form a strip across the top of the brain from ear to ear. These signals are then sent down to the midbrain, brainstem, and spinal cord to create motor movements.
The movements create new sensory input, which then cycles back up the same loops. The entire nervous system is highly recursive, and we can conceptualize 'thought' and signals ping-ponging back and forth from the sensory association areas to the motor association areas and back.
Since this has already become very long, I will stop here. I hope this is helpful.
Interesting article. The human brain has evolved in line with specific natural laws to help us think of ways to adapt and prosper in numerous circumstances. Two fundamental features of the universe are order and chaos. People recognized these natural early on and explained them with the Yin and Yang in the East and the book of Genesis in the West.
The left brain evolved to aid orderly thinking - math, logic, etc. While the right handles chaos - art, uncovering new insights through epiphanies, etc.
Sorry, it appears my posts are appearing in reverse order. Please read the third post below first. Then the second, and this post last.
Organizing principle #3: the right brain processes bottom-up, the left top-down.
I am in total conflict with you here. Both sides of the brain use both bottom-up *and* top-down processing. The function of the nervous system is to allow the organism to respond to the environment. In the bottom-up/top-down paradigm, sensory input is bottom-up processed as signals travel to and within the brain, then the brain top-down processes the signals to send back to muscles to cause behavior. As mentioned in #2, the sensory signals cross to the opposite side of the brain for the more concrete levels of processing. So input from the right side of the body is processed in the left brain. The left brain then sends signals to the right side of the body for movement, and vice versa.
The closest version of bottom-up/top-down is the division of the brain into sensory (bottom-up) areas in the parietal lobes (above and behind the ear) and motor (top-down) areas in the frontal lobes (above and between the eyes and ears).
One has to be careful about bottom-up/top-down conceptualization, though because it implies a locus of initiation. The nervous system is constantly processing in both directions. If there is a starting point, it is the sensory input from the environment.
As to left-right differences, as I discussed in my original post, the major difference appears to be language processing in the left. But these areas are highly connected with areas on the right (like server farms on different continents).
Organizing principle #2: collect things at similar levels of abstraction.
#2A: the rear of the brain is concrete, the front is abstract.
Yes to #2 (subject to discussion of ‘near’ in #1, and if you are careful about the meaning of ‘abstraction’.) Information in processed significantly even in areas outside the brain. For example, the retina of the eye has already identified the edges of objects before the signal leaves the eye to travel to the brain.
Mostly no to #2A. The fact that the rearmost part of the brain, the occipital area, is more ‘concrete’ is an accident of the placement of the special senses of vision. For some reason the nervous system tends to place association (abstraction) areas on the opposite side of the body from the sensory input they process. Touch on the right side of the body is processed by the left brain. Visual input from the front of the body goes to back of the brain.
However, the *dominant* organization of ‘concrete’ and ‘abstract’ is the center of the cortex (a strip across the top from ear to ear) being the most concrete. As you move away from that strip, both forwards and backwards, the brain gets more abstract.
But just because they are far apart doesn’t mean the abstract areas in the front of the brain are less connected to the abstract areas in the back. An analogy is that the server farms of Google are spread across the world, but the connections among the server farms are arguably stronger than the connection of a particular server farm to a house a mile away.
I will address your hypotheses more directly. On my reading you have three main hypotheses, or organizing principles, related to brain function. I will address each in a separate response.
Organizing principle #1: related functions are more connected to each other.
Yes, but more connected doesn’t necessarily mean *near* each other. One general principle of the nervous system is that there are at least 3 different types of connection. The most ‘dense’ and slowest is where the axon (the output) of the neurons connect to many nearby neurons. Signals spread in a wavelike fashion.
A second connection is where signals are collected from several thousand neurons and sent to other, relatively nearby units. The cortex of the brain is formed of 6 layers of sandwiched neurons, some layers are more ‘local’ in function and often form a column of related neurons. Others layers communicate with other columns.
A clumsy, but serviceable analogy, is that each neuron is a microprocessor, the locally-related neurons in a column are a server, and an area of the cortex encompassing several thousand columns is a server farm.
Finally, the third connection type is fast connections among ‘server farms’, that can span the entire brain. The connections of the corpus callosum, which crosses from left to right brain and back, are of these third type.
You gave a lot of detail, but didn't relate it much to the hypothesis I offered. Do you see this detail as confirming, conflicting, or not relevant?
"That was poor phrasing by me, sorry."
No, your generic you was clear ... Hanson's just a dick with a fragile ego.
"On what basis do you claim that I'm unaware of details?"
Sigh.
https://ase.tufts.edu/cogst...
Disclaimer: say no to ad hominems. Oh, no, I'm not saying that you, Robin Hanson, are unaware of details. That was poor phrasing by me, sorry. I tried to show that human brain (mostly young areas, such as the frontal lobes developed in historical time of behavioral modernity) as an information processing entity can recognize/formalize only tiny fraction of what's going on behind the gates of the ACC. And this tiny fraction of accessible information that we, humans, can study is an end result of several interrelated neurological ensembles. There is evident separation, of course. An much more than just binary separation. So, our conscious self employs young and articulate frontal cortex to interpret those signals from our ancient parts of the brain. That's the point. Better now.--In no way I was trying to overwhelm you with details (I suck at writing and don't really know how to explain brain anatomy without mentioning certain things). Nor did I try to prove something here, on your blog. Just wanted to say hi, question some alternatives that might be complementary to the binary concept. I have no further comment.
I trust you're aware of the details, but whether your theory successfully explains them can't be evaluated by this post. [Not to carp, but some such disclaimer might have been useful.]
[The view that the left brain is deductive and the right inductive is fairly common. But that this represents the basic principle of its organization apparently novel.]
While Jo Mo doesn't state his objections crisply, some can be inferred. One such objection (which I thought of too) is that the capacity of the brain to reorganize itself may be in tension with the conclusion that the design principle is based on keeping things separate. (A testable inference might be that when the brain reorganizes functions from one side to the other the organism will experience more self-sustaining credit cycles than one with a brain not reorganized. An unexpected consequence of minor brain damage.)
I of the Vortex by Rodolfo Llinas (2002) says as much as can be usefully said about the core function of the brain. From his premise we can begin to understand why our large brains can be subject to useless fantasizing such as speculations without utility.
On what basis do you claim that I'm unaware of details? That I didn't give a long list of details like you did? You think blog posts and their comments should be contests to show who can list more details?
That's the point. You can contemplate on general patterns (top-down) while simultaneously being unaware of a lot of gritty details underneath those patterns (bottom-up). A very smart design that reduces a lot of noise.
Data: Meditation increases connections. Assumption of mine: Advanced meditators display symptoms of what you call self-supporting cycles. Plausible cause of self-supporting cycles: increased excitatory connections, not inhibitory
The claim that a big function of the connections between hemisphere is to prevent self-supporting cycles is consistent with that total connection size depending on many factors.
"The main risk of mixing credit directions is creating self-supporting credit cycles not well connected to real needs. This may be why the connections between the two hemispheres are mostly inhibitory, reducing activity." If that were true, what would it suggest about increased connectivity between hemispheres in meditators? Assuming meditation involves learning to perceive sound as sound (not windy leaves of a tree, not bus, etc.)? Seems to suggest that excitatory connections are being formed, or perhaps your hypothesis tends towards being wrong Related: https://www.ncbi.nlm.nih.go...https://www.ncbi.nlm.nih.go...
I appreciate your writings. If you are going to further pursue brain structure and function, I hope you will be careful about your sources. There is a lot of magical thinking about brain function, especially the right/left differences.
I have done research in neurophysiology, and part of my function when I was at NIH was to keep up with the neurobiology research. Most of my experience, however, has been more clinically oriented with patients with neurological dysfunctions.
If you look at the big picture of neurological function, the system works to receive sensory input, process that input, and create functional (mostly motor) output that enhances the organisms response to the environment. If you stimulate one side of a worm, it's nervous system will contract muscles that withdraw that side away from the noxious stimulus. The functions are preserved in humans, and generally operate at the spinal cord level.
As nervous systems evolved and became more complex, new functions have been layered on top of the old ones (I hesitate to use the term pathways, because they lead people to conceptualize wires, or the old style phone systems; the functions are more comparable to distributed communication like in the Internet.) The organisms reactions to sensory input have become much more complex, and separated from the input in both time and space. A high level of separation has often been labeled as abstraction.
There is a high level of processing throughout the nervous system. Humans have 100+ billion neurons, each neuron functions as a processor with several hundred to several thousand inputs (from dendritic connections to upstream neurons) creating a continuously variable output (axonal firing rate) that signals downstream neurons.
At the risk of error through over-simplification, the function of the spinal cord is the first few levels of processing of sensory input from the body and creating simple movements (a cat's hind limbs can make stereotypical walking movements even with a totally isolated spinal cord). This already highly processed information passes to the brain stem, which adds in specialized senses such as taste and hearing, and basic visual input. The brain stem output is mostly homeostatic functions of the body (blood pressure, breathing, sleep, etc.).
The midbrain adds in more complex 'rehearsed' movements, e.g., a batter hitting a baseball. Some types of memory (e.g. procedural) exist at the midbrain level (and if you have a broad definition of memory, even lower in the nervous system). The cortex is the most 'abstract', and appears to have evolved to add flexibility and improved pattern recognition across time. A fairly accessible example of the functions of the cortex is to compare cats and dogs. Cats have smaller a simpler cortex. While they are superb at highly stereotypical behaviors, the system can often be 'fooled' again and again, e.g., a cat will pursue a laser pointer dot, as it triggers the 'prey-pursuit' functions. The more complex cortex in a dog allows more specific training, more 'sociability', etc.
The human cortex has two main input areas which are somatotopically arranged: the visual system (in the back of the brain, the occiput), and sense of touch/position (a strip arching across the top of the brain from ear to ear). These sensory inputs then spread into sensory association areas, which, as you posit, become more 'abstract' the farther you get from the primary sensory areas. This spread travels 'forward' from the visual areas, but 'backward' from the somatosensory (touch) areas. These 'parietal' areas includes the back half of the brain. The areas 'furthest' from the primary sensory areas are the temporal lobes, which extend laterally and forward from the parietal lobes. The temporal lobes contain the most abstract functions such as language.
Up to this point, most of the brain functions are the same on both sides. The major difference in right and left brains is in the functions of the temporal lobes, with the left side usually containing most verbal language function, and the right containing non-verbal language and non-language communication functions. If you look at the detail of studies on right/left brain differences, the ones ascribed to left dominance are those with tests that require verbal language (using/remembering words). Although we thing of communication as being verbal, over 90% of human communication is non-word based, and over 60% is non-verbal. For example, think of the difference in texting versus a phone call. Texting loses all the non-word communication such as voice tone, speed, pitch, etc. Now look at the increase in communication that comes from face-to-face interaction, where non-verbal information such as facial expressions, posture, etc. are added. Damage to the right brain causes loss of skills in those non-verbal domains.
The output from all of these sensory association areas (including the temporal lobes) transits forward to the motor association areas of the frontal lobes, which formulate behaviors at various levels of complexity. The most 'abstract' motor areas are in the front, and the information flow moves 'backward' to the primary motor areas, which are just forward of the primary somatosensory area and similarly form a strip across the top of the brain from ear to ear. These signals are then sent down to the midbrain, brainstem, and spinal cord to create motor movements.
The movements create new sensory input, which then cycles back up the same loops. The entire nervous system is highly recursive, and we can conceptualize 'thought' and signals ping-ponging back and forth from the sensory association areas to the motor association areas and back.
Since this has already become very long, I will stop here. I hope this is helpful.
Interesting article. The human brain has evolved in line with specific natural laws to help us think of ways to adapt and prosper in numerous circumstances. Two fundamental features of the universe are order and chaos. People recognized these natural early on and explained them with the Yin and Yang in the East and the book of Genesis in the West.
The left brain evolved to aid orderly thinking - math, logic, etc. While the right handles chaos - art, uncovering new insights through epiphanies, etc.
I explore more of these natural laws on my blog.
Check out www.simplyurban.org