Increasing my ignorance on Neural Science
I’ve been working so much this past year that I haven’t been able to keep up with my blogging. My wife sent me on a two day mancation just before Xmas to unwind before the family chaos began. I have a new text book I bought a while back that I’ve been dying to read and just haven’t had time titled; “Principles of neural science” 5th edition, that I brought along with me.
The pace of new discoveries in neural science has been staggering in recent years and difficult to keep up with. These days I do a lot of work for science and research institutions attempting to simulate physics on massive new scales. Converting science into large scale computer simulations is my new favorite enterprise but the irony is that I spend so much time building the computer frameworks for these giant simulations I scarcely have time to keep up with the thing I love most about doing it… the science itself. I’ll talk about that exciting work in a future blog when the projects are public but I’m currently working on some of the largest science simulations in human history, really cool stuff. Fundamentally it all boils down to trying to figure out how to simulate life itself. I sequestered myself for a few days with this massive text book, a copy of “Quantum Mechanics for dummies” and a tomb of HP Lovecraft stories just to round out the mancation reading mix. *No I don’t need to learn introductory Quantum Mechanics, it’s a handy reference for the core equations and mathematical tools. I didn’t permit myself to bring a laptop or Mathematica with me on mancation so I had to resort to old fashioned pen and paper mathematics.
As usual I didn’t get far into my new text book (which is fantastic) without running into a few of the old neural science canards that have long troubled me. It’s really trendy out in the valley and in academic circles to talk about artificial intelligence and brain simulation as though we are on the cusp of major advances. We’re REALLY not. This book is full of amazing new information and insights about the field and yet I always find myself looking for a few magic words. There’s nothing I like better than finding the phrase “The purpose of which is still unknown”, or “the function of which is still a mystery” in a 1700 page treatise on the state of Neural Science knowledge. Whenever I find these phrases I realize that despite holding 1700 pages of “knowledge” on the subject of Neural Science, the book is missing several thousand more pages that will one day be devoted to the quickly dismissed subjects addressed by those phrases today. Specifically there are answers to questions about simulating life that I’m looking for when I read these things and the answers are still missing… Not only are the answers missing, they’re the answers to questions fundamental and vital to having any chance of making progress at simulating life and thought with a computer. I have my opinions about what these answers may look like one day when they are properly studied, but for the the moment, nobody seems to know.
Take a sentence like this;
“The functions of (glia) astrocytes are still mysterious. It is generally thought that astrocytes are not essential for information processing.”
Preceded by the statement:
“There are 2 to 10 times more glia than neurons in the vertebrate central nervous system.”
…so… neural cells called glia make up over 66% of the brain. We don’t know what they do but…
“Although glial cells do not generate action potentials, they have recently been found to participate in neuron-glial signalling processes. The significance of this signaling is still poorly understood, but it MAY actively help regulate synapse development and function.”
So two thirds of the neural cells in the brain we don’t understand but everybody is REALLY CONFIDENT that we’re close to modeling complete human minds and making progress at artificial intelligence using neural networks that don’t include models for glial cells. Hey I’m a computer scientist… what do I know? Let’s look for some recent literature on glial cells just to see what we think we know about them.
“As development proceeds, the brain becomes larger and the primary mode of neuronal migration from the VZ changes. Because of the greater distances, neurons require what was originally identified as a special population of cells within the VZ called “radial glial guides” to support their migration (Rakic 1972). Much like neurons migrating via somal translocation, radial glial guides extend a basal process that attaches to the pial surface of the brain. However, the nucleus of the radial glial cells remains in the VZ, and the basal process forms a kind of scaffolding along which neurons can migrate (see Fig. 8b). The migrating neurons attach themselves to the radial glial guide and move along the cellular scaffold out into the developing cortical plate (Nadarajah and Parnavelas 2002). Each glial scaffold can support the migration of many neurons. Although the radial glial guides were originally thought to be a special, transient population of cells, it has recently been discovered that the cells that provide the scaffolding are actually the neural progenitor cells (Noctor et al. 2001; Noctor et al. 2002; Parnavelas et al. 2002; Weissman et al. 2003).”
Oh… they’re just essential to BUILDING the F**KING BRAIN! Think about the implication of this. The brain is obviously not built by itself, it’s built by a bunch of witless cells powered by genetic machinery with no neural network to think or guide them. The brain is assembled by cells with no brains! These things are the little mechanics the wire up the human mind in order for it to function correctly! Who cares what a neural network can do, I want to know how cells can build computers that think! Other than that how important can glial cells be to brain function once the damn thing has been constructed?
“As we shall see in later chapters, chemical synapses are functionally and anatomically modified through experience and learning as much as during early development.”
Read a little further and glial cells actively participate in re-wiring the mature brain in response to learning! So the brain can’t learn or work correctly without these mysterious cells making decisions (somehow) about when it is time to physically rewire your neural-network while it is running! This is the kind of stuff that persuades me that we aren’t even remotely close to understanding or modeling how a mind works. What am I going to learn from reading the rest of this book if it can’t answer questions like this? So on the same page that we get a lesson on how mysterious astrocytes (glial cells) are we get a diagram like this. Here we have our mysterious astrocyte connected to blood vessels, the neuron itself AND mediating the output signal of the axon with the explanation that; “astrocytes are thought to have a nutritive function”. This makes my head hurt. There is clearly a lot more going on in this cellular relationship than a nutritive one. Even a nutritive function to a neuron is also a messaging relationship. How would the brain know what was going on with the human circulatory system if it didn’t have some way to “listen” to our blood?
Let’s look at this another way. Plants have no brains, no nervous systems, no complex organs and no circulatory system, yet they have behaviors. They behave differently at different times of day, under different weather conditions, some of them even catch and eat insects in real-time, with no brain. Plants also learn. They can adapt to changes in their immediate environments without brains or nervous systems. Clearly, long before animals came along, living organisms had passive mechanisms for “thinking” and “learning” that didn’t rely on having a nervous system. The human body and mind still has those systems in place… or we wouldn’t be able to survive to construct our own brains. The brain almost certainly also has to have an intimate communication relationship at the molecular signaling level with these primal forms of thought and information processing. It’s implausible that our minds manage to “think” and regulate the human body hormonaly without being able to communicate in the new fast animal way, electrically, AND in the old slow plant way, chemically. This astrocyte sure looks like it’s interfacing a neuron with the circulatory system. So what might an astrocyte (glial cell) be listening to the circulatory system for?
“Immune responses in the CNS are common, despite its perception as a site of immune privilege. These responses can be mediated by resident microglia and astrocytes, which are innate immune cells without direct counterparts in the periphery. Furthermore, CNS immune reactions often take place in virtual isolation from the innate/adaptive immune interplay that characterizes peripheral immunity. However, microglia and astrocytes also engage in significant cross-talk with CNS-infiltrating T cells and other components of the innate immune system. Here we review the cellular and molecular basis of innate immunity in the CNS and discuss what is known about how outcomes of these interactions can lead to resolution of infection, neurodegeneration, or neural repair depending on the context.”
Please recall from high-school biology class that our immune system operates predominantly through our circulatory system. In short there are whole new levels of brain function interacting with our circulatory system that we don’t even remotely comprehend yet. At least 66% of the brain is dedicated to essential computing stuff we’re just beginning to be aware of and it doesn’t work even remotely like a neural network. In fact neural networks in the brain are so massively mediated by these glial cells it’s really not even clear that our neural networks are even cheap approximations for how the brain actually operates.
Okay enough about the “mysterious” glial cells that probably don’t do anything important. Let’s learn about the neurons and the neural networky stuff that goes on in the brain that we seem to feel really confident and knowledgeable about. It appears that there are six types of neurons. Five types of neurons that have different ways of handling input and output. The function of each type seems to make great sense as I read through all of their functions. The sixth type of neuron… can you spot it? The Interneuron has no outputs… What? What use is a computing element that only takes input and produces no output? I’m skipping to that chapter because THAT is the interesting neuron type.
“Local interneurons have short axons and form circuits with nearby neurons to analyze small pieces of information. Relay interneurons have long axons and connect circuits of neurons in one region of the brain with those in other regions. The interaction between interneurons allow the brain to perform complex functions such as learning, and decision-making.”
20%-30% of the neurons in the brain… that’s the thinking bit… are Interneurons.
“Accordingly, the 6-layered neocortex, as the center of the highest nervous functions such as conscious perception or cognition, has the largest number of interneuron types”
Let’s try to put some perspective on this. The other five neuron types all appear to be well understood, make sense, are largely hard wired from birth and form the basis for our models of computational neural networking. They are boring, easily simulated and most importantly don’t participate much in actual thought. They mostly worry about keeping your heart beating and helping you reach for potato chips successfully. The Interneuron, which you almost never hear or read about, doesn’t have axons like the other neurons and uses an entirely different neuro-signaling paradigm for local communications with other interneurons which is largely inhibitory. Imagine a crowded room full of people all telling one another to shut-up so they can hear what somebody else is saying. We don’t understand how this alternative model for neural computing works BUT we believe it gives rise to consciousness and memory.
“The term interneuron hides a great diversity of structural and functional types of cells. In fact, it is not yet possible to say how many different kinds of interneurons are present in the human brain. Certainly hundreds; perhaps many more.”
It turns out that the name interneuron isn’t even a defined type of neuron, it’s a basket name for a family of neuron types with an unknown number of members and unknown variability. So just to be clear, we have no model for the kind of neural communication that gives rise to consciousness and we don’t even have a way to categorize the number of cell types that participate in it yet. I’m feeling more educated by the page.
By chapter 3 I have run smack into the modern bio-sciences dogma on mutation. So in the year 2016, 5th edition text book on Principals of Neural Science the DEFINITION of mutation is…
“Although DNA replication generally is carried out with high fidelity, spontaneous ERRORS called mutations do occur. Mutations can result from DAMAGE to the purine and pyrimidine bases in DNA, MISTAKES during the DNA replication process, and recombination that occur during meiosis.
The rate of spontaneously occurring mutations is low but measurable, spontaneous mutations make a significant contribution to human genetic DISEASE. The frequency of mutations greatly increases when the organism is exposed to chemical mutagens or ionizing radiation during experiment genetic studies.”
Mutations are accidents, they’re errors that cause disease in the presence of toxic environmental conditions. I have come to believe that this nearly ubiquitous view on mutations is the single largest mental block modern scientists have to understanding life and thought. I’ve written about it previously but it never ceases to amaze me how strongly and confidently academics talk about mutations in these terms. I think that at this point the scientific evidence that mutations are a highly controlled and systematic biological process essential to the operation of our brains, immune system and reproduction is overwhelming. In my previous blogs on this subject I’ve cited several fascinating research studies on the subject, but the basis of my conviction that this view of mutation is simply wrong and medieval is based on my understanding of computing. Living organisms self-program. The human brain self-programs. Our DNA doesn’t contain a blue print for our bodies or our minds, it contains rules and guidelines that billions of cells working in concert and in communication with one another use to assemble us in the presence of wildly varying environmental and resource conditions. The cells have to innovate, they have to adapt, they have to MUTATE in order to build and maintain a working organism in the presence of constantly varying environmental circumstances. Deliberate and systematic genetic mutation is a tool actively employed by our genomes to get us up and running.
What is one of the most “toxic” mutagenic chemical compounds known to science? Nitric Oxide… a natural by product of cellular respiration. Wouldn’t you tend to think that if mutation was a BAD thing from the cells point of view that it would have evolved a really strong resistance to being mutated by its own respiratory by-products? When do cells produce more NO than usual? When they are under stress. So a cells response to encountering an environmental condition it can’t cope with is to try mutating itself. When scientists test mutagens they just bathe the subject cells in them and see what happens which not surprisingly usually produces a useless or destructive mutation result. However in the few studies that have been performed in which they subject cells to stress and let them self-mutate, the result, shockingly tends to be BENEFICIAL mutations.
The prevailing view that genetic mutations are “errors” that cause “diseases” that need cures seems to have blinded scientists to a fundamental mechanism that may give rise to the most important and powerful properties of life and thought that we are struggling to answer. We can’t make software that programs itself and evolves in any kind of generalized systematic way, living machines appear to have the property of self-design and systematic mutation is a powerful answer to how such a feat is achieved. In the same chapter this academic view towards mutation runs amok with the following example.
“Autism is a common, devastating developmental DISORDER characterized by deficiencies in language acquisition, difficulties in social interactions, and stereotyped interests.”
So we’re about to get a lesson on how genetic mutations cause the Autism disorder right?
“There is considerable variability between autistic individuals. Children affected with autism have a higher frequency of seizures and cognitive problems than the general population, and some are severely disabled. However, many autistic individuals have normal or above-normal intelligence, and can, with appropriate care, go on to lead highly successful lives.”
…wait for it…
“Autisim has a very strong heritable component which should increase the CHANCES of identifying the underlying genes.”
WHAT! You don’t even KNOW what the underlying genes are yet and are using AUTISM as the example in the chapter of the text book on Neural Science titled GENES AND BEHAVIOR!!!!
“Autism is a common DISORDER, yet there is no single genetic location for autism as there is for much rarer Williams syndrome. At least one part of the explanation for this difference is that autism is not really one disorder caused by a single kind of genetic LESION, but rather a cluster of related disorders caused by a variety of different genetic changes.”
In other words “Austism” is not actually a disease or a physical condition anybody can identify. It’s not linked to any gene or group of gene mutations. It’s like the Interneuron of diseases, just a basket disease for a bunch of stuff we don’t understand yet that may have no common correlation. People who don’t think and function the way WE think they should must have a disease so let’s put all the socially odd people in the Autism disease category. We have no physiological medical basis for calling this a disease but let’s diagnose half a percent of the population as having this condition. What if the reason they can’t quantify it as an actual physical disease is because it’s just an arbitrarily defined set of personality traits that falls within the perfectly natural range of human personalities and behaviors. Being gay also used to be considered a disorder. *They still can’t find a gay gene either.
So why is my text book chapter about how MUTANT genes can influence behavior teaching me about personality “disorders” that have no genetic basis? Very confusing.
“The observation that common DISEASES such as autism are less well understood than rare diseases such as Rett syndrome applies broadly to psychiatric disorders. blah.. blah.. blah.. it has been difficult to identify the genes associated with many chronic human psychiatric DISEASES like chronic depression, and anxiety disorders, which affect a substantial fraction of people.”
Do you see the premise being established here? COMMON diseases, those are the ones we’re all going to be diagnosed as having, must eventually be discovered to be the result of genetic LESIONS that require study and TREATMENT because they are by definition bad and ERRORS in our genomes. If you are anxious, depressed, anti-social, violent or stupid, we’ll eventually find the genetic lesion to blame for your condition and be able to prescribe a treatment for it. But what if these “psychiatric diseases” are the result of the normal healthy operation of our genomes? What if these genetic lesions, even if they are found to exist are simply part of how the natural range of human personality is expressed by a complex combination of genetic and neurological processes? If these behaviors were really all “unhealthy” shouldn’t evolution have scrubbed them away millions of years ago? What if it’s not our genomes but the modern environments we live in that make these formally beneficial personality traits seem adverse?
As a computer scientist with an interest in computationally simulating living systems what these guys call “mutations” and “diseases” looks like essential biological machinery for answering questions like why can human beings write original code and design complex machines to solve arbitrary problems, not just problems our DNA was coded to cope with. The more I read these text books the more I feel like an archaeologist studying ancient Greek science texts explaining how the sun and planets revolve around the Earth. It’s not that I know better than they do, just that it’s clear they are wrong and actively glossing over huge bodies of ignorance to focus on the subjects they think they understand. In just a few chapters I find that I need a glossary for sciency sounding terms that mean “I don’t know”. Ancient map makers used to label uncharted regions of the world “Here there be monsters”, the question is did the sailors know what that meant?
Interneuron: A family of neural cells of unknown size making up 20%-30% of the brain probably responsible for conscious thought and memory that we have no idea how to characterize or explain their communication modes.
Glial Cells: A family of neural cells that make up at least 66% of the brain that are probably responsible for building the damn thing and rewiring the brain physically in response to learning but otherwise don’t perform any important function and can safely be ignored if you accept that they are just passive scaffolding.
Mutation: Something that is almost always random, bad and in need of medical treatment. Mutations are the cause of any disease we assert that you have. Even if we haven’t identified them yet we eventually will.
Autism: A vague set of personality traits that 1 in 200 of our children have that have no medical or genetic basis for, but they’re weird people, so we’re sure we’ll find the genetic mutation responsible for their disorder eventually.