This is such a fascinating subject. I've often pondered this little thought experiment. _"Describe _*_exactly_*_ the shape and size of a human femur. Your instructions can be as long as you like, but you may only use four letters. Go!"_
I love to picture our cell structure as an orchestra or a choir, pretty much everyone in the orchestra knows what to sing and how to harmonize, they know their melodic role in the piece, and usually you get a group of players/singers play/sing same lines and some times harmonize basic to complex melodies that evoke different emotions in us listeners. If a player/singer falls out of time or skips some notes, the whole orchestra will lift them up and eventually they will know their place again and play their melody the one needs to be heard. So, the encoded anatomy is more like a song that's being sung by the whole community of the cells in our body, and once the song misses some notes the new born cells will jump in to fill their role and sing their life melody. Imagine an old folklore song that you recall but you don't really remember the whole lyric, if you try to sing it with friends that share the memory with you, as a group you will eventually sing the whole of it correctly.
Shape and function is information; where is the information stored? The question perhaps is how can proteins (made from the nucleus) effect the electromagnetic signals that direct shape and function? Also, what feedback mechanism informs the nucleus to shift protein generation.
One thought i havent heard u touch on is how much can cells copy. Why do most embrios grow inside the mother. The cells might be using input from mother to know how to build
So after watching this video I read through a lot of the comments. I then went and watched some more in-depth stuff of his at greater length. My background is computers and physics and I have dabbled in cell and molecular biology and Neuroscience. I cannot help but think about Quantum superposition and entanglement. It has been demonstrated that there are quantum effects going on in the brains of mammals. It must also be true for the way that the protein structures fold. It would answer the question as to how they do it so quickly. Add to that the holographic theory of memory. And how much information is stored in a hologram. And the astronomical number of permutations that are present to encode information in DNA. I would not be surprised at all if the electrical patterns that we are seeing or encoding the shape information in a manner of Quantum superposition
Fascinating talk ended before I was ready! As a software engineer I was trying to imagine what kind of program could produce a "self-healing" result from simple inputs that update along the way. Off the top of my head, it looks like a main routine that spawns and passes most functions to subroutines that deal with the finer and finer details (but that stay in touch with other routines above and below). If I were younger....
ML and similar researchers are, of course, building upon previous work, with immense collaboration and sharing occurring on a global basis, leading to the exponential increases in knowledge and astonishing technological development that we’re seeing. Meanwhile, during the quest to eradicate many diseases of humans, improve human lives, nonhuman life is being destroyed, the biosphere is being degraded, fished-out oceans polluted w/ plastic, resources depleted, and 1,000,000 new humans added q 4-5 days. So the exhaustion of the Earth’s resources by too many human parasites, the scary but inevitable military applications of AI, necessitated by an escalation in war and conflict over dwindling resources, and the unintended consequences of bio technological/genetic engineering will make Dr. Levin’s fine work superfluous, as catastrophes loom on the horizon.
This is such a powerful way to think about biology. And the examples are excellent. The nephron example at 8:45 clearly shows that there is a system with memory of the setpoint structure and access to multiple mechanisms to effect the changes required to become that shape. It also shows that the system is getting feedback from the growing structure. It must know when it is working with cells large enough that only one cell is required to make a tube with sufficient size lumen. They are such big ideas that people brains literally balk when I try and tell them about it. I hope short videos will make it easier for people to be able to digest the new concepts.
I always wondered how your cells know how to grow and correct errors in 3d space like how do they define a boundary in space so they don't just keep growing. That made me think of cancerous tumors and how they seem to have lost the ability to stay within their spatial domain as they grow out of control.
they lose growth regulating genes due to dna damage. by 'default' singular cells emerged to just multiply successfully as soon as possible. then regulation of this process evolves to be more adaptive and survive as a colony [see volvex]. these growth-suppression genes get slammed out of the chain and we get faster multiplication. hope your T-killers detect this asap.
The book Life Becoming (I think it's called) lays out a bunch of mechanisms. Stuff like: when the density of certain signalling cells get too low, or weird algorithms where the rate by which the cell creates or destroys certain rigid structures equalizes. It's analogous to how hair reaches an equilibrium point based on how quickly it falls out and how quickly it grows. I only vaguely remember, but it's cool.
I would suspect that cells constantly monitor their environment. So cells that are on the edge of an embryo, the surface cells, are constantly broadcasting surface signalers. Cells that are next to them broadcast a unique signaler and so on and so forth. By monitoring the average gradient of signaling proteins a cell can get a rough idea of where it is in the embryo as it develops. Then based on their locations cells then attempt to begin to differentiate. My guess is that they start putting out signalers that recruit neighboring cells that recruit neighboring cells until the signal strength reaches a critical threshold at which point an election process begins amongst them to further differentiate, activating some parts of the genome and shutting down others. So say you have an embryonic liver. One cell decides its going to be a liver cell and it tells the surrounding cells to also become liver cells. If those cells are undifferentiated they recruit and turn off their ability to be anything and instead become liver cells, signaling what they are doing all the while. When the signal reaches a strength threshold the cells stop recruiting and start dividing. Cells that are on the boundary layer know this because they dont have liver cells on all sides, only on one side, so signaling proteins would be roughly half of what a cell inside the group might experience. This allows the liver cells to output signaling proteins "Hey I am on the boundary layer!" to other cells. Then once the boundary is established you can link all the boundary cells together with mechanical linkages and monitor those linkages for damage. A loss of a connection could change how the tensile strength of the boundary cells gets interpreted. So if you have mechanical changes, plus damage signals in the water, the cells know to interpret it as damage and so begin the healing process. Whereas if you just have flexing alone, then that triggers nothing. Basically an AND gate if I were to convert it into my field (I work in IT. So feel free to call my meanderings a load of baloney!) I think how the system works is that each cell, having the same genome as every other cell, uses hormones and signals to know which bits to turn on and which bits to turn off. This defines which proteins and signals and chemicals a cell produces vs which ones it doesnt produce, and the presence of these proteins are what chemically and mechanically define how the cells interact with each other. With billions of chemical pathways inside a cell acting almost like very very very tiny logic gates, the cell is able to make decisions based off of its immediate surroundings. Taken in aggregate all these chemical logic circuits combine to produce emergent behavior. The reason that say, Picasso always produces a frog... You have cells that elect to be the front of the organism and cells that elect to be the middle and cells that elect to be the back end. Then you have cells that are eyes and they are basically telling all their neighbors they are eye cells. Except those cells chemically do not respond to eye cell signals. So instead the eye cells look for the signals that indicate front vs back, and begin to migrate to the front. There is probably a mechanism to allow them to migrate around. As the cells migrate they detect 'Face' signals which tells them its going in the right direction. Eventually they find the cells they are supposed to be next to and stop there. They overcorrect because as this is a chemical biological process, they can get too strong a signal and move too enthusiastically, whereupon they encounter boundary cells where they arent supposed to and move away from that boundary layer signal. Because there are TRILLIONS of possible protein combinations, you can theoretically have as many signals as you need to set off as many chemical pathways as you need for these logic circuits that then chemically set off yet more pathways that allow the cell to move around mechanically. Its why you first have a mass of undifferentiated cells in a ball because thats the easiest shape that has a defined inside and outside. At least that is my layperson thinking. No idea how close to the truth I actually got.
@@thomashenry4798 Thank you so much for the very detailed comment! I suggest you join our online community to further share your thoughts with SEMF members: semf.org.es/participate/join.html
@@SEMF I appreciate your reply and your content to be clear continue the content... Your content is a meeting place of different ideas , all with the focus of cross pollination ... allowing the absorbing of and discussion of and thus development of deeper understanding for everyone involved ... truly EPIC and heart warming In your about section you boil that down to the "inclusion of diverse viewpoints" which has come to represent of those groups and people who use such language as actually meaning we value all viewpoints but only the "valid ones" and that validity is not created by nor supported by the diverse discussion but rather limits it. As I feel your truly about multi angled and multi disciplinary discussion then I feel you need to be more clear about that in your language... unless you are trying to capitulate to that "jargon" which unfortunately signals fundamental corruption of your stated goals (ideals) In any case. Take it as you will
This is a much needed video. I want to hear much more about this question. I'd like to understand as a lay person, and not a genetics and development scientist
@@realGBx64 Interesting! Well thanks for the feedback, it is always useful to gather impressions. The tree-like shape is a remnant of our former logo that had the "Tree of Knowledge/Science" as a centrepiece. The current design is intended to evoke dendrites/neurons, particle traces, etc. If you want to learn more about the society behind this channel you can visit our website: semf.org.es/mission/
Wow, top notch. It’s cool to see someone talk with specific knowledge about computer science and biology. It seems like the video ended without the question really being answered, or perhaps I just struggled to understand it. Somehow the tissues seem to ‘know’ what morphology they are meant to become, even if they need to take a different approach to getting there. Do you have a mechanism in mind to explain this?
i think we must ask the question WHAT is encoded before "how" a blueprint encodes a shape and produces identical bolts and nuts. whereas frogs are all different, different IN SHAPE, yet they are all "similar" what IS this "similarity"? let alone Trees -- unique as effing fingerprints! the idea with geometrical space distortion is very misleading, i'd start from the CONNECTIVITY, certanly a connectivity "scheme" of a body must be much more invariant than the shape.
I've been wondering about this for years this for years. Where is the blueprint? How do cells know where they are at in the tissue/organ/organ system/creature? How do they know when to divide or stop dividing? This is exciting stuff!!
I recall watching a channel abiut single cell organisms and sometimes, and not nearly rarely, you see something that makes you think these things are intelligent & perfectly capable of some sort of reasoning (at least on some level)
Not only cells have intelligence but i think DNA also has intelligence because it has goal directed behavior.some components of the cell like ribosome,mitochondria also have goal directed behavior.Is it possible that cellular components also have intelligence?
Oh wow, this is very insightful. I am wondering how this take on closed loop pattern homeostasis, with a certain notion of "agency" and "intelligence", might be applicable to the patterns of (gross and fine) neuroanatomy and morphobehavioral outcomes in cognition. One system I have been thinking about for some time is the phonological component of linguistic competence. I am intrigued that all normal children will develop some kind of "morphospace" of phonemic distinctions (a system of "cognizable types"), incorporating the particular phonetic acoustic signals they happen to be exposed to in their household and community. What is surprising, and indicates that there are multiple paths to "the same" kind of outcome space is what happens to deaf children (or even hearing kids), especially those that are raised in a household or community where a lexically rich and fully grammatical sign language is used. (These are known as Deaf of Deaf kids, or in the case of hearing kids, Children of Deaf Adults or CODAs, and they grow up as highly fluent native signers.) They acquire an "emic system" that classifies "etic" gestures (distinguishing the phonemic-like level of cognizable types from the phonetic level of percept-level or subcognitive sensory signaling). My working hypothesis is that the similarity of behavioral outcomes (and potentially some of the encoding in brain morphology) can be understood in terms of the topological invariants on the geometric space of what seems to correspond to your notion of morphospace. Specifically, it seems like the etic space (of acoustic or gestural signals) undergoes some kind of (strong) deformation retract to a subspace of isomorphisms (identity functions or identifiability "functions" or perhaps homotopy types). There is a dramatic reduction in the degrees of freedom, and the phonemic (or gestural emic) types of the household-community are the outcome, different for each speech community or signing community but remarkably similar in structure when viewed as mathematical invariants. These ideas about phonology are about a relatively small subsystem of linguistic cognition, sort of higher percepts, that are closely tied to more elaborate linguistic systems like lexis and grammar constructions and semantics and pragmatics (and other areas of cognition like concept formation and recognition of intentional mental states in others). I have some ideas how a more general framework of cognizable types can accommodate some of the other subsystems of language, but phonology seems a practical place to start. This fits in with a broader framework of "information flow" influenced by the work of Barwise and Seligman (a 1997 book with that title), which may have a general enough mathematical model of information to accommodate the notions of biological information you are developing. I need to start looking into some of the papers you cite.
@@SEMF You are welcome. I am trying to schedule binge-watching some of your content, hopefully I will have something more to say in the near future. Cheers!
@@fbkintanar That's great to hear! If you want to contribute to the community, you are very welcome to join our mailing list and Discord server: semf.org.es/participate/join.html
You say something that is 'goal' oriented rather than simply 'emergent' is uncomfortable for biologists, but I would argue it is indeed the opposite because as you've described, the reality is that these things ARE in fact goal oriented. So biologists should be very interested in this because it reflects what IS happening, rather than someone's preconceived ideology about how they think the world 'should' work.
Did you check if the wave function collapse algorithm is a good approximation for this? It seems like it can solve very similar problems by only looking at local state like cells can.
This really was an amazing contribution. We are having a Levin-focused event in our community. You can join the live discussions on our community here: semf.org.es/participate/join.html
Your channel, and this video in particular caught my attention due to a completely different reasons - i was planning on making a tool for generating beings for a video game, that have anatomy and can maintain homeostasis of some kind. The things you said gave me a lot of ideas. I doubt it will be useful for advanced academic purposes, but i bet it's going to be a fun little toy, that might or might not give some intuition about further stuff. I reckon feedback loops play a huge role - a homeostasis can be seen as a chaos control mechanism selectively developed by nature, to maintain optimal parameters of a chaotic system throughout its lifetime. Chaos control is currently an emergent tech that has many useful practial AI implementations, and in some sense can reverse-engineer (approximate) some natural mechanisms. So, designing and compiling a living system would inevitably be the designing and compiling a system of feedback mechanisms for chaos control.
This "goal directed behavior" is so surprising to us because it seems to fly in the face of the unconsciously assumed reductionism that permeates our culture. The whole, in many cases, may not only be "greater than the sum of its parts", but also more fundamental.
Thanks for your interest! You can participate in the SEMF community by joining our network (free) following this link: semf.org.es/participate/join.html You will get access to our community forums, mailing lists and early announcements about future events for registration.
Do we have some clear experimental example of when the cells fail to form the typical structure they usually form, after some disturbance? If yes, then I think researchers should focus on that because it would allow us to pinpoint a necessary mechanism that didn't work. It's like how solving mazes is easier if you start from the end. Just observing that cells know how to rearrange themselves (on multiple levels and following multiple paths) simply looks like "magic" and doesn't really tell us how. Also, perhaps the complexity of how DNA translates to such diverse mechanisms can be better explored by automation. Imagine if we could automate making genetic (and epigenetic) changes and observing the effect on millions of test cell groups, it would become feasible then to launch a massive brute force search to pinpoint the genes necessary for this regeneration and be able to observe a huge number of variations and understand the effect of each change...
is studying the limit of regeneration useful to figure out where anatomical information is stored? how far can we slice an axolotl until it stops regenerating correctly?
Just want to say how wonderful it is to have credentialed scientist asking such a question. If we assume that information is encoded at the level of base pairs then the total amount of information that can be represented by a single molecule of human DNA is about 1.5GB - so not very much at all. If you were to model the human body down to the level of the individual cells (so not down to protein level) and tried to do that in a 3D modelling software like Blender then you'd need literally TB of data. So Mr. Levin right to ask how the 'missing' information is encoded. I think answering this question will open a door to a new panaroma for biology, and probably physics as well. Bravo!
What makes living systems different from nonliving is simple: Order. The degree order imposed on matter and the flow of energy through it is specified by digitally encoded symbolic data, AND an operating system of immense complexity which regulates the production of each nanomachine and the regulation of each metabolic and structural component in patterns where groups of machine and processes unfold to produce a generative whole. A blueprint may be used to build a factory. But in living systems, instead, there is a written instruction manual made of words, interpreted by a cooperative robotic system which interacts with software which "understands" how to translate those words into structures and further instructions,, with regulations in timing. If the first group of builders builds the first room specified, copies his textts but highlights what the next builder should do and shouldn't do, and this process continues. What should make this extraordinary, actually a miracle, is these processes have built in error correcting systems which can accommodate changing environments. The individual words which are written for the builder to read may be "door" but the where and when of installation cannot be found in a simple sentence.
Great insight to understand life! But sill the question “where it is encoded?” is unanswered. I believe that species pass through generations not only physical characters but also experiences. I mean not only instincts, but also complex behaviour patterns (e.g. cuckoo chick ejects eggs out of the nest). The question is relevant not only to morphology, so what is the answer? Kind of The Little Prince’s magic box answer: somewhere within the very first cell AND/OR somewhere outside the cell. The part “outside the cell” cannot be neglected, because “the law of physics,” at least at atom level (and sub-atom) provide infrastructure life is based on. Can we reject the idea of Cloud storage? Or, may be single cell encodes much more than we thought
Thank you so much for your interest! We are glad you are enjoying our content. Please visit our website for more information on the society behind this channel: semf.org.es/
This was an update in the general context of the concept of Spatiality in our conference. Arguably a summary of previous, previously known work, but very relevant to this concept of space and shape.
I liked the talk, but rather than suggesting routes of future research, it was describing why the question is a question to begin with & that it is yet to be explained. To be fair it does end by suggesting it’s unlikely purely emergent, but (based on the title) I was expecting some potential explanations of how it is encoded & more about ongoing research into those explanations.
That's a fair description. Michaels work is amazing and very promising but still early stages. We look forward to inviting him and his collaborators again!
so one starts out as a point in morphospace (I like that term) and gradually expands, stretches and shifts to occupy different regions. sort of like mthematical functions in a vector space? can we then work out the orthonormal equivalent in cell mrphology?
What makes you think you can back-derive the genes from the anatomy? Certainly there no unique 1:1 function, at least in the backwards direction. Anatomy is a high-order effect of self-interacting generation, therefore there are many initial conditions that will produce the same result.
@@SEMF I thought that was the leading question for the 'anatomy compiler' part. He says you *should* be able to draw the anatomy and infer the stimuli for genes to become that anatomy. Perhaps later he says that is silly or something I'm not sure.
It is Memory, which is present all over your body, not only in the Brain, the rest is done by food intake. If you eat a piece of bread it becomes you, if another person takes a bite from that bread it becomes his body with all features according to the memory collected.
I recently wondered about this so thanks youtube. Being a c++ coder I was wondering where the memory of the huge design of our body is situated. I googled a bit and I found that we know at least some genes that code for some structure. Good to know that overall its still a mystery. Being in addition a mathematician I know that a fractal can be defined by a simple formula. Together with this feedback idea we can at least get some intuition.
@@SEMF so fractals exist in state space where consecutive points have indeterminate or chaotic vectors. That is the N+1 point is indeterminate given variation of initial conditions, but the attractor, or pattern remains fairly consistent within a set of mean values. These values could be represented by tensor, or sets of n vectors with probabilistic values affected by a recursive function. Like you indicated in the video. It might look like magic, but a recursive function, acting through state space would be like the complex plane.
DNA might describe a cell but may also convey rules to each cell like cellular automata does for each pixel. Maybe there's biochemical markers and cellular sense making of its environment.
I wonder if an individual cell can detect how many cells away is the nearest cell of a different tissue. like if a skin cell can detect how many skin cells are between it and the nearest subcutaneous fat cell, or if a liver cell can detect how many liver cells are between it and the nearest blood vessel cell, etc. Because if a cell *can* tell these things, then it seems to me like it would be a lot easier to figure out how anatomy emerges; each cell could roughly "triangulate" its position relative to other tissues in the body, and reproduce or die as needed to maintain an appropriate range of distances from the appropriate surrounding tissues. This is just a hunch but I'd be interested to hear what you think of it @SEMF
This is Faraday and Maxwell level science. Their explorations in electromagnetism led to electric motors, telegraphs, phones, computers, etc. What will Levin's explorations into the bioelectric role in morphogenesis bring?
History reveals the weaponisation and marketisation of science by invested insiders seek prior advantage of insider knowledge, kept secret or surrounded by obfuscations. The Field approach is NOT a control model but a mindset of control seeks according to its own conditioned thinking, under the mask of serving humanity, for possession and control. Hence The Biotech revolution of augmented or captive human to such systems is close at hand. Living waters are structured by nanoscale quantum charge domains as a medium of translation of information unfolding from a whole. The pathological model rises from a sense of lack, exclusion, conflict, denial, set to defences that boost and reinforce the model as the basis for interpretation. Wholeness is a balancing alignment of a life as contextual expression. But All the king’s horses and all the king’s men cant see it as that predicates on a broken life that must be fixed.
Ok. Simple model problem. Suppose we want to generate a ball of cells with a certain number of cells as its radius or a certain distance as radius. What would be a simple way to achieve that. What would each cell have to measure, and what would be a mechanism to compare that to a number stored in for example the genome? A ball seemed like the simplest Morpheus space I could imagine, maybe there are simpler once.
Simpler problems like these are probably important to investigate. Perhaps you would like to share your thoughts within our community: semf.org.es/participate/join.html
Perhaps DNA contains the blueprint and specs for making whatever needs to made by simply storing pointers. The pointers point to a specific blueprint stored in some huge universal database from which the cells somehow retrieve the required information. For example, Pointer 6582240352982549 references the geometry of my big toe. Questions remain: Where exactly is the database, and how is the data retrieved? If you can answer either of these, I will gladly share my Nobel Prize with you.
I ve been in medicine field since many years ago, and I can call it after more than 33 years of practice , that God is real. !!! Congratulations for your incredible videos!!!!❤❤❤
So if a salamander was born with a deformed leg, that gets cut off, does it grow back a deformed leg or does it fix itself given another chance to grow the leg.
Awesome!!! I always wondered about how shape and symmetry were “encoded”. Since I was younger I noticed that people who have extra fingers usually are not symmetrical and if symmetry was encoded in the DNA a mutation causing an extra finger should produce it on both hands, at least there were two sets of genes to control ontogenesis of each body side the body, which does not make much sense.
Morphology would have to be this way, because there’s too many unknown influences to build by sequence. “Shapes” will be encoded as modes (not static targets). I would guess it’s encoded on the cell level. Default mode would just be split, and other modes would be based on signals. To find the encoded “shapes”, you would have to locate all mechanisms that respond to signals of any kind, in a reasonably reliable way. Forms would be encoded as a byproduct of a massive mode mode hierarchy that determines each cells current role in the total system. If all cells are essentially the same, but specialize based on signals, to encode new forms, it’s all about the mode hierarchy. The mode hierarchy is going to be insanely huge, so identifying the levels and branches will be the main challenge. Once identified, individual sections of it can be reprogrammed with the appropriate signals applied to the appropriate levels. But preventing reversion would be a challenge when trying to alter a grown body.
@@SEMF Thank you for your videos, Micheal is one of the most brilliant people I’ve heard speak. Since the mode-like cell behavior is not crazy, here is my guess for where the forms are stored. Assuming evolution is lazy/cheap, the easiest/cheapest way to encode physical functions (including “target form seekers”) would be to use the structure or “shape” of the control surfaces (the structures in cells that behave differently based on inputs; I don’t know the proper terms). This is explained at the end. Reading DNA is probably expensive, so you must store the real-time access information somewhere else. Putting all this together means you would expect DNA to encode instructions for an ultra general purpose biological machine that can copy itself, stay connected with, authenticate, and communicate with the copies, and change between a vast landscape machine types (resulting in a vaster landscape of possible behavior spaces), depending on its own state, signals from copies, and perhaps signals from non-copy “friendly” cells. You don’t design a body, you fine-tune a single super-cell that can copy itself and take on an infinite number of different roles/modes to fill its space in the environment. The cheapest way to store real-time morphology information (returning to the explanation) would be in the shape of the control surface, because its physical structure can be such that it reaches a natural balance for free. The analog thermostat example is appropriate. If the mercury has risen to the midpoint due to the temperature, you may ask “where is the form of 1/2 stored in the thermostat?” But of course it’s not, it’s encoded in the design of the control surface of the thermostat, which functions based on the environment it exists in. DNA would encode instructions for building a cell-building framework (some kind of super-proto-cell), that would encode or re-encode a specific morphology in its control surfaces based on what type of cell it is, based on signals from other cells and it’s own state. Since the real-time form/morphology information is encoded in the physical structure, it would not appear like data, even though its structure is data driven. At no point would you find anything that represents a body part shape because it would actually be represented by the control surface shape that seeks the proper steady state, for its purpose, at its location, for the purpose of resulting in the desired overall morphological shape (even though it has no knowledge of this grand purpose). The actual forms we see when looking at bodies would be meaningless side effects from the cells perspective. So, a person who is taller will have “taller” control surfaces. Not literally taller, but they will be constructed differently in a way that results in a steady state that results in a taller person. The control surface differences that encode real-time form information just by existing, could be revealed with machine learning by comparing them in detail (if detailed active-cell snapshots are possible) to the large scale morphologies, with enough properly tagged sample sets from one species. Cells would have to somehow periodically do an expensive “check” of the DNA to make sure the control surfaces are still shaped correctly, otherwise bodies would drift off into strange shapes as they take on damage, and quickly stop working (because of the hierarchical/recursive nature of the signal-response chains). You should also expect some kind of checksum or redundancy somewhere in this process. If true, altering DNA would be the most difficult way to manipulate forms, because you’re programming under several layers of abstraction. But it would be the most permanent. If you tried to manipulate the control surfaces, the form would revert after DNA checks. However, I suspect the surfaces are far more complex than this simplistic picture, and that there are infinite ways of manipulating the cell systems to capture new forms permanently by hacking the existing systems (coaxing the system into new steady states by pushing specific signals past expected boundaries; essentially causing overflow errors). Again, if true, every cell should know how to do *everything* that any other cell of the same species can do, if properly coaxed through the mode hierarchy.
The answer may lie in a new understanding of physics rather than biology. We humans are natural-born engineers. As such, we model after machines not only isolated, naturally occurring systems, but also the basic laws of physics, sharing with machines a local-evolution-of-state `grammar'. However, there is a strong case to be made that it is this mechanistic paradigm which is to blame for the stagnation on many of the stubborn open problems in physics. The implications for biology are precisely what Michael is eluding to: `goal oriented' evolution in the sense that the goal is already `out there' - in the 4D block universe.
@@SEMF You can find a concrete proposal titled "Questioning the mechanistic-universe paradigm using chaotic systems" on my ResearchGate page (I would have included a link had TH-cam allowed). Still a preprint. Not an easy task finding a journal willing to publish such heresy...
@@yonatan2806 Thanks very much! Please feel free to join our SEMF online community and share your preprints and ideas here: semf.org.es/participate/join.html
Perhaps the nephrons forming the lumen are paying attention to the surface tension across their cytoskeletons. If multiple cells could bend together and achieve their desired surface tension amount together, then great. If only one giant polyploid cell is available, then that's fine too.
The key is the distribution of regulatory substances in the body and their interaction with information/rules in the cell: if there is too much or too little of the substance, less or more will be produced. If you have rules for multiple substances, you get a stabile pattern (standing wave), like the stripes of a zebra, that define areas of the body. Think how this works on a scale with hundreds of regulators that create a complex pattern to define aspects of the morphology. If this system is perturbed, if will strive to get back into balance. If a new cell migrates into a tissue, its role is defined by the mix of regulators in its surroundings. The informational model for this would be a 3D pattern of Turing engines, interacting with each other locally, writing and reading streams of 'tape'. Each engine has hard-wired ROM data (DNA) as well as local RAM storage (RNA, methylizations). With the right kind of wiring and initial settings, the machines will interact to create an organism. If the settings are wrong, the organism will not converge and die off. Nature selects the working settings.
The answer to this question can be found by deeply understanding the mechanism of self-similar structure transformation (Fractals). Just as the lungs were mapped by Mandelbrot himself so too can anatomical structures (Mandelbrot BB. The Fractal Geometry of Nature. W. H. Freeman & Co. Ltd; 1982.). Just as the Mandelbrot set forms 3d acorns, so to do organs and anatomy form finite shapes and they dont "stop" they continue but they iterate on themselves until no more energy is left in the system in which they are derived from. The key would be to use cross sections view (such as from fMRI etc) of development and with OpenCV to devise the baseline "fractal" that creates said shape. As to "where" this information is actually encoded, that answer becomes more clear when you start deconstructing all biology in terms of its baseline fractal. The encoding of this information is not present in the system directly, just as if we were to look at ChatGPT and view one of the transformations it wouldn't make sense as to how that singular transformation is able to perfectly structure sentences and the things it does. The information is encoded more in a gestalt way. (Fractals of Brain, Fractals of Mind: In Search of a Symmetry Bond (Advances in Consciousness Research, No 7)
Indeed! That's a very interesting angle! Many SEMF community members would love to discuss these ideas further. We invite you to join by following this link: semf.org.es/participate/join.html
There’s a game called cell lab where you make little multicellular organisms by defining a set of cell types, for each saying the angles of the descendants upon splitting, and what cell types the two descendants are. Combine this with the physics based precession of cells pushing and deforming each other and you arrive at some morphology. I’d say that encoding is not a great word for organisms that we know in real life, as code suggests it’s all layed out in readable form, when really it’s just a lot of evolutionary biases and constraints from physics, that is with the exception of the more organised aspects of dna, which aren’t the full picture.
That sounds super interesting! Why don't you share some links or info about that game with our community? You can join at no cost other than filling a quick form here: semf.org.es/participate/join.html
Congratulations ! It was not easy to summarize such a complex issue. I am not sure this riddle will ever be solved, not for technical reasons but idological ones . The denial of some kind of finalistic intelligent inherent to natural mechanisms make possible to go beyond mere description of facts.
I thought we already knew this. It’s not just life, but physics as well. Universal darwinism applies to all self-stabilizing systems ie any dissipative structure whatsoever - including physics itself. Clearly complexity and order supervenes on self-interaction at various levels of scale - it is not built into the initial conditions like your reversible, idealized, linear, physics would have you believe. Most phenomena are non-linear open dynamical systems, so if they are not autopoietic or at least self-stabilizing, they will dissipate due to the stochastic nature of reality, and therefore will not perpetuate long enough to converge onto self-replication and eventually become an ideal Quine.
Clearly. that anatomical information has to be in the genome, as the animal starts from a single cell. But that information is not in the DNA sequence. Sequence is a low-level language for making proteins. This language includes sequence of amino acids, but also meta information like starting and stopping points, where to snip, and regulatory areas. There must be a higher level language layered on top of the DNA sequence. This language must be capable of describing anatomical structures by shape, size, and timing. What are the components of this language. What are it's equivalent of Words and Syntax? Or does it use some entirely different method of encoding information? I seems completely invisible to us. Why? Is it so completely different from the way we think?
When we consider Evolution, how does that interplay here? I was always taught that evolution means mutations in genes which lead to a change in fitness. But if the morphological change is independent of genes, then that approach is wrong. So what exactly is evolution. Does that mean we have no clue how it works. If I'm not mistaken evolutions main supporting fact is that they compare the morphological tree based on fossils and the phylogenetic tree and make them fit.
Michael's work just points to spatial patterns of bioelectricity as other possible forms of information encoding, similar and complementary to genetic encoding.
For me the key has to be information. We know that DNA is the primary information repository. Biology abhors inefficient processes, we know that only 1% of DNA codes for proteins. Therefore it seems obvious to me that the 99% of non coding DNA must be serving another function and is highly likely be the key to intra and inter cellular organisation through, as yet unknown mechanisms. It must be independent of neurological mechanisms or flora would be unable to regenerate form.
@@SEMFgood luck, I worked on cytokines. Even simple regenerative processes are horribly complex in higher eukaryotes and failed. Best advice is to choose as simple a model as possible and the simplest process you can find. We are missing fundamentals, which is why I’d start with looking for additional information storage mechanisms which have to be in plain sight.
@@SEMF We wouldn’t carry that huge “junk” load of conserved information. That 99% of “non coding” sequences remain largely unstudied. My wild speculation as to how it’s read and the outputs would be unsupported by data If I were wanting to understand the mechanisms, the logical place to start are one or both of the gametes, all the necessary information must already be present prior to the zygote. The one exception could be if there were a maternal transfer of information whilst in the uterus, it would be easy to exclude this by looking at developmental issues in animals grown fully ecto uterine, which I believe can be now be done. Not my field of understanding though.
Perhaps thoughts are nothing but the experience of these feedback loops from a future ideal state to the present state, the present state incurring the form of the ideal future state in order to maintain or attain homeostasis towards the shape of the "body" the thought is pointing towards....on the higher level of self-organization of the human body, these are the higher dimensional possible/probablistic topologies of a resonant electric surface that the neurons of the brain create to mimic the internalized world picture it has in relation to these short lived ideal edifices of that they spontaneously create aka thoughts
Wow, this has to be one of the best videos on TH-cam.
Levin Lab's work really is amazing!
This is such a fascinating subject. I've often pondered this little thought experiment. _"Describe _*_exactly_*_ the shape and size of a human femur. Your instructions can be as long as you like, but you may only use four letters. Go!"_
Great thought experiment!
Spoken like a poet. Nice.
I love to picture our cell structure as an orchestra or a choir, pretty much everyone in the orchestra knows what to sing and how to harmonize, they know their melodic role in the piece, and usually you get a group of players/singers play/sing same lines and some times harmonize basic to complex melodies that evoke different emotions in us listeners. If a player/singer falls out of time or skips some notes, the whole orchestra will lift them up and eventually they will know their place again and play their melody the one needs to be heard.
So, the encoded anatomy is more like a song that's being sung by the whole community of the cells in our body, and once the song misses some notes the new born cells will jump in to fill their role and sing their life melody.
Imagine an old folklore song that you recall but you don't really remember the whole lyric, if you try to sing it with friends that share the memory with you, as a group you will eventually sing the whole of it correctly.
That's a beautiful elaboration!
Shape and function is information; where is the information stored? The question perhaps is how can proteins (made from the nucleus) effect the electromagnetic signals that direct shape and function? Also, what feedback mechanism informs the nucleus to shift protein generation.
Great question!
One thought i havent heard u touch on is how much can cells copy. Why do most embrios grow inside the mother. The cells might be using input from mother to know how to build
That seems to be one of the most characteristic functions of cells indeed.
So after watching this video I read through a lot of the comments. I then went and watched some more in-depth stuff of his at greater length. My background is computers and physics and I have dabbled in cell and molecular biology and Neuroscience. I cannot help but think about Quantum superposition and entanglement. It has been demonstrated that there are quantum effects going on in the brains of mammals. It must also be true for the way that the protein structures fold. It would answer the question as to how they do it so quickly. Add to that the holographic theory of memory. And how much information is stored in a hologram. And the astronomical number of permutations that are present to encode information in DNA. I would not be surprised at all if the electrical patterns that we are seeing or encoding the shape information in a manner of Quantum superposition
There is a real possibility that quantum entanglement is involved in the process. Great comment!
Fascinating talk ended before I was ready! As a software engineer I was trying to imagine what kind of program could produce a "self-healing" result from simple inputs that update along the way. Off the top of my head, it looks like a main routine that spawns and passes most functions to subroutines that deal with the finer and finer details (but that stay in touch with other routines above and below). If I were younger....
If you want to hear more from him, he had a great interview on Lex Fridman's podcast
this is an extract from the longer th-cam.com/video/jLiHLDrOTW8/w-d-xo.html
Michael is so brilliant, thank you for this !
It was a pleasure to host him at SEMF!
It's so crazy this is the only man talking about this stuff.
He is representative of the work at his lab and by his collaborators, perhaps he is the most visible but not the only one?
@@SEMF Yeah I'm sure that in the community there are more people, but in mainstream I think this is getting nearly zero attention.
Pure genius; this is paradigm shifting stuff!
Most likely so!
ML and similar researchers are, of course, building upon previous work, with immense collaboration and sharing occurring on a global basis, leading to the exponential increases in knowledge and astonishing technological development that we’re seeing.
Meanwhile, during the quest to eradicate many diseases of humans, improve human lives, nonhuman life is being destroyed, the biosphere is being degraded, fished-out oceans polluted w/ plastic, resources depleted, and 1,000,000 new humans added q 4-5 days.
So the exhaustion of the Earth’s resources by too many human parasites, the scary but inevitable military applications of AI, necessitated by an escalation in war and conflict over dwindling resources, and the unintended consequences of bio technological/genetic engineering will make Dr. Levin’s fine work superfluous, as catastrophes loom on the horizon.
@@commanderthorkilj.amundsen3426 This is an important notice!
@@commanderthorkilj.amundsen3426 Blimey! Doomer, much?
This is such a powerful way to think about biology. And the examples are excellent. The nephron example at 8:45 clearly shows that there is a system with memory of the setpoint structure and access to multiple mechanisms to effect the changes required to become that shape. It also shows that the system is getting feedback from the growing structure. It must know when it is working with cells large enough that only one cell is required to make a tube with sufficient size lumen.
They are such big ideas that people brains literally balk when I try and tell them about it. I hope short videos will make it easier for people to be able to digest the new concepts.
Thank you very much for the comment! We are also hoping shorter videos will be useful!
This man has a nobel prize in his bags. It's only a matter of time.
Very likely the case!
I always wondered how your cells know how to grow and correct errors in 3d space like how do they define a boundary in space so they don't just keep growing. That made me think of cancerous tumors and how they seem to have lost the ability to stay within their spatial domain as they grow out of control.
they lose growth regulating genes due to dna damage.
by 'default' singular cells emerged to just multiply successfully as soon as possible.
then regulation of this process evolves to be more adaptive and survive as a colony [see volvex].
these growth-suppression genes get slammed out of the chain and we get faster multiplication.
hope your T-killers detect this asap.
The book Life Becoming (I think it's called) lays out a bunch of mechanisms. Stuff like: when the density of certain signalling cells get too low, or weird algorithms where the rate by which the cell creates or destroys certain rigid structures equalizes. It's analogous to how hair reaches an equilibrium point based on how quickly it falls out and how quickly it grows. I only vaguely remember, but it's cool.
Michael's research is indeed a great advance in this direction!
I would suspect that cells constantly monitor their environment. So cells that are on the edge of an embryo, the surface cells, are constantly broadcasting surface signalers. Cells that are next to them broadcast a unique signaler and so on and so forth. By monitoring the average gradient of signaling proteins a cell can get a rough idea of where it is in the embryo as it develops. Then based on their locations cells then attempt to begin to differentiate. My guess is that they start putting out signalers that recruit neighboring cells that recruit neighboring cells until the signal strength reaches a critical threshold at which point an election process begins amongst them to further differentiate, activating some parts of the genome and shutting down others.
So say you have an embryonic liver. One cell decides its going to be a liver cell and it tells the surrounding cells to also become liver cells. If those cells are undifferentiated they recruit and turn off their ability to be anything and instead become liver cells, signaling what they are doing all the while. When the signal reaches a strength threshold the cells stop recruiting and start dividing. Cells that are on the boundary layer know this because they dont have liver cells on all sides, only on one side, so signaling proteins would be roughly half of what a cell inside the group might experience. This allows the liver cells to output signaling proteins "Hey I am on the boundary layer!" to other cells. Then once the boundary is established you can link all the boundary cells together with mechanical linkages and monitor those linkages for damage. A loss of a connection could change how the tensile strength of the boundary cells gets interpreted. So if you have mechanical changes, plus damage signals in the water, the cells know to interpret it as damage and so begin the healing process. Whereas if you just have flexing alone, then that triggers nothing.
Basically an AND gate if I were to convert it into my field (I work in IT. So feel free to call my meanderings a load of baloney!)
I think how the system works is that each cell, having the same genome as every other cell, uses hormones and signals to know which bits to turn on and which bits to turn off. This defines which proteins and signals and chemicals a cell produces vs which ones it doesnt produce, and the presence of these proteins are what chemically and mechanically define how the cells interact with each other. With billions of chemical pathways inside a cell acting almost like very very very tiny logic gates, the cell is able to make decisions based off of its immediate surroundings.
Taken in aggregate all these chemical logic circuits combine to produce emergent behavior.
The reason that say, Picasso always produces a frog...
You have cells that elect to be the front of the organism and cells that elect to be the middle and cells that elect to be the back end. Then you have cells that are eyes and they are basically telling all their neighbors they are eye cells. Except those cells chemically do not respond to eye cell signals. So instead the eye cells look for the signals that indicate front vs back, and begin to migrate to the front. There is probably a mechanism to allow them to migrate around. As the cells migrate they detect 'Face' signals which tells them its going in the right direction. Eventually they find the cells they are supposed to be next to and stop there. They overcorrect because as this is a chemical biological process, they can get too strong a signal and move too enthusiastically, whereupon they encounter boundary cells where they arent supposed to and move away from that boundary layer signal.
Because there are TRILLIONS of possible protein combinations, you can theoretically have as many signals as you need to set off as many chemical pathways as you need for these logic circuits that then chemically set off yet more pathways that allow the cell to move around mechanically.
Its why you first have a mass of undifferentiated cells in a ball because thats the easiest shape that has a defined inside and outside.
At least that is my layperson thinking. No idea how close to the truth I actually got.
@@thomashenry4798 Thank you so much for the very detailed comment! I suggest you join our online community to further share your thoughts with SEMF members: semf.org.es/participate/join.html
“How could we have a navigation system that can have goals in anatomical space?” Living systems are wonderful. Very nice talk. Thanks.
Glad to see you enjoyed our content!
@@SEMF how can one claim science but actually be a proponent of the opposite of science ??? Please tell us
@@xiaonanw6374 What do you mean?
@@SEMF I appreciate your reply and your content to be clear continue the content...
Your content is a meeting place of different ideas , all with the focus of cross pollination ... allowing the absorbing of and discussion of and thus development of deeper understanding for everyone involved ... truly EPIC and heart warming
In your about section you boil that down to the "inclusion of diverse viewpoints" which has come to represent of those groups and people who use such language as actually meaning we value all viewpoints but only the "valid ones" and that validity is not created by nor supported by the diverse discussion but rather limits it.
As I feel your truly about multi angled and multi disciplinary discussion then I feel you need to be more clear about that in your language... unless you are trying to capitulate to that "jargon" which unfortunately signals fundamental corruption of your stated goals (ideals)
In any case. Take it as you will
@@xiaonanw6374 Thanks for the kind words!
Holy crap that was mind-blowing!
Especially the "Picasso-Tadpole" 11:37
That was a really incredible example!
This is a much needed video. I want to hear much more about this question. I'd like to understand as a lay person, and not a genetics and development scientist
Look forward to future SEMF events, we will be having many opportunities to learn from the basics. semf.org.es/participate/join.html
Fascinating research. What a great project to be part of.
It truly is fascinating!
Fascinating new perspectvive
It really is!
Thank you for putting this on TH-cam
Very interesting concept!
Would bone remodeling of a fracture site be an example of this process??
It definitely looks like so!
This channel and this video in particular are excellent finds. Thanks!
Thank you very much for your kind comment. We feel privileged to be able to bring this cutting-edge research to a large audience.
Wow.Mind blowing.
It really is!
This talk described the threshold of science and magic
It really did!
Look up, Rupert Sheldrake - Morphic Resonance and Morphic Fields
@@SEMF "inclusion and diversity " wtf man
@@xiaonanw6374 huh?
@@generichuman_ their description in about the channel..
Wow, based on the channel logo I was not expecting a proper science lecture. Well done!
Thanks for the feedback! If we may ask, what about the logo made you think that the content wouldn't be a "proper science lecture".
@@SEMF looks like a spiritual tree of life logo of some cult to me.
@@realGBx64 Interesting! Well thanks for the feedback, it is always useful to gather impressions. The tree-like shape is a remnant of our former logo that had the "Tree of Knowledge/Science" as a centrepiece. The current design is intended to evoke dendrites/neurons, particle traces, etc. If you want to learn more about the society behind this channel you can visit our website: semf.org.es/mission/
Wow, top notch. It’s cool to see someone talk with specific knowledge about computer science and biology. It seems like the video ended without the question really being answered, or perhaps I just struggled to understand it. Somehow the tissues seem to ‘know’ what morphology they are meant to become, even if they need to take a different approach to getting there. Do you have a mechanism in mind to explain this?
i think we must ask the question WHAT is encoded before "how"
a blueprint encodes a shape and produces identical bolts and nuts.
whereas frogs are all different, different IN SHAPE, yet they are all "similar"
what IS this "similarity"?
let alone Trees -- unique as effing fingerprints!
the idea with geometrical space distortion is very misleading, i'd start from the CONNECTIVITY, certanly a connectivity "scheme" of a body must be much more invariant than the shape.
This video is just a small extract from the full talk, see video description
I've been wondering about this for years this for years. Where is the blueprint? How do cells know where they are at in the tissue/organ/organ system/creature? How do they know when to divide or stop dividing? This is exciting stuff!!
This is super exciting indeed!
I recall watching a channel abiut single cell organisms and sometimes, and not nearly rarely, you see something that makes you think these things are intelligent & perfectly capable of some sort of reasoning (at least on some level)
@@osaimola Interesting!
They don't know, there is no global/top down organization. Local rules evolve into equilibrium
@@johnk6757 Local rules evolving into equilibrium is indeed a very important abstraction to model the kind of phenomena Michael Levin studies.
Not only cells have intelligence but i think DNA also has intelligence because it has goal directed behavior.some components of the cell like ribosome,mitochondria also have goal directed behavior.Is it possible that cellular components also have intelligence?
Intelligence in the sense described here, absolutely! That's a very real possibility.
Oh wow, this is very insightful. I am wondering how this take on closed loop pattern homeostasis, with a certain notion of "agency" and "intelligence", might be applicable to the patterns of (gross and fine) neuroanatomy and morphobehavioral outcomes in cognition. One system I have been thinking about for some time is the phonological component of linguistic competence. I am intrigued that all normal children will develop some kind of "morphospace" of phonemic distinctions (a system of "cognizable types"), incorporating the particular phonetic acoustic signals they happen to be exposed to in their household and community.
What is surprising, and indicates that there are multiple paths to "the same" kind of outcome space is what happens to deaf children (or even hearing kids), especially those that are raised in a household or community where a lexically rich and fully grammatical sign language is used. (These are known as Deaf of Deaf kids, or in the case of hearing kids, Children of Deaf Adults or CODAs, and they grow up as highly fluent native signers.) They acquire an "emic system" that classifies "etic" gestures (distinguishing the phonemic-like level of cognizable types from the phonetic level of percept-level or subcognitive sensory signaling). My working hypothesis is that the similarity of behavioral outcomes (and potentially some of the encoding in brain morphology) can be understood in terms of the topological invariants on the geometric space of what seems to correspond to your notion of morphospace. Specifically, it seems like the etic space (of acoustic or gestural signals) undergoes some kind of (strong) deformation retract to a subspace of isomorphisms (identity functions or identifiability "functions" or perhaps homotopy types). There is a dramatic reduction in the degrees of freedom, and the phonemic (or gestural emic) types of the household-community are the outcome, different for each speech community or signing community but remarkably similar in structure when viewed as mathematical invariants.
These ideas about phonology are about a relatively small subsystem of linguistic cognition, sort of higher percepts, that are closely tied to more elaborate linguistic systems like lexis and grammar constructions and semantics and pragmatics (and other areas of cognition like concept formation and recognition of intentional mental states in others). I have some ideas how a more general framework of cognizable types can accommodate some of the other subsystems of language, but phonology seems a practical place to start. This fits in with a broader framework of "information flow" influenced by the work of Barwise and Seligman (a 1997 book with that title), which may have a general enough mathematical model of information to accommodate the notions of biological information you are developing. I need to start looking into some of the papers you cite.
Thank you for such a well-written and informative comment!
@@SEMF You are welcome. I am trying to schedule binge-watching some of your content, hopefully I will have something more to say in the near future. Cheers!
@@fbkintanar That's great to hear! If you want to contribute to the community, you are very welcome to join our mailing list and Discord server: semf.org.es/participate/join.html
You say something that is 'goal' oriented rather than simply 'emergent' is uncomfortable for biologists, but I would argue it is indeed the opposite because as you've described, the reality is that these things ARE in fact goal oriented. So biologists should be very interested in this because it reflects what IS happening, rather than someone's preconceived ideology about how they think the world 'should' work.
How are goals defined in nature?
I agree! He’s got a Nobel prize waiting for him
Very possibly!
Mind blowing! Excellent presentation. Thanks.
We thought so too! Thanks for your comment!
Did you check if the wave function collapse algorithm is a good approximation for this?
It seems like it can solve very similar problems by only looking at local state like cells can.
This really was an amazing contribution. We are having a Levin-focused event in our community. You can join the live discussions on our community here:
semf.org.es/participate/join.html
Your channel, and this video in particular caught my attention due to a completely different reasons - i was planning on making a tool for generating beings for a video game, that have anatomy and can maintain homeostasis of some kind. The things you said gave me a lot of ideas. I doubt it will be useful for advanced academic purposes, but i bet it's going to be a fun little toy, that might or might not give some intuition about further stuff. I reckon feedback loops play a huge role - a homeostasis can be seen as a chaos control mechanism selectively developed by nature, to maintain optimal parameters of a chaotic system throughout its lifetime. Chaos control is currently an emergent tech that has many useful practial AI implementations, and in some sense can reverse-engineer (approximate) some natural mechanisms.
So, designing and compiling a living system would inevitably be the designing and compiling a system of feedback mechanisms for chaos control.
That sounds fascinating! We invite you to share such projects with our community: semf.org.es/participate/join.html
This "goal directed behavior" is so surprising to us because it seems to fly in the face of the unconsciously assumed reductionism that permeates our culture. The whole, in many cases, may not only be "greater than the sum of its parts", but also more fundamental.
This is a very interesting observation and one that will be addressed in future SEMF events and discussions.
I wish I could participate. This is so fascinating.
Thanks for your interest! You can participate in the SEMF community by joining our network (free) following this link: semf.org.es/participate/join.html
You will get access to our community forums, mailing lists and early announcements about future events for registration.
This is a really interesting question extremely well presented
We thought so too. Michael is a very clear speaker!
Thank you guys, keep up the good work!
Thank you, it is our pleasure!
what is it about amphibians that makes them regenerate more easily?
Do we have some clear experimental example of when the cells fail to form the typical structure they usually form, after some disturbance? If yes, then I think researchers should focus on that because it would allow us to pinpoint a necessary mechanism that didn't work. It's like how solving mazes is easier if you start from the end. Just observing that cells know how to rearrange themselves (on multiple levels and following multiple paths) simply looks like "magic" and doesn't really tell us how.
Also, perhaps the complexity of how DNA translates to such diverse mechanisms can be better explored by automation. Imagine if we could automate making genetic (and epigenetic) changes and observing the effect on millions of test cell groups, it would become feasible then to launch a massive brute force search to pinpoint the genes necessary for this regeneration and be able to observe a huge number of variations and understand the effect of each change...
Levin's work shows some direct experimental evidence. You can check other videos in our channel for it!
is studying the limit of regeneration useful to figure out where anatomical information is stored? how far can we slice an axolotl until it stops regenerating correctly?
Just want to say how wonderful it is to have credentialed scientist asking such a question. If we assume that information is encoded at the level of base pairs then the total amount of information that can be represented by a single molecule of human DNA is about 1.5GB - so not very much at all. If you were to model the human body down to the level of the individual cells (so not down to protein level) and tried to do that in a 3D modelling software like Blender then you'd need literally TB of data. So Mr. Levin right to ask how the 'missing' information is encoded. I think answering this question will open a door to a new panaroma for biology, and probably physics as well. Bravo!
That's a very relevant observation!
What makes living systems different from nonliving is simple: Order. The degree order imposed on matter and the flow of energy through it is specified by digitally encoded symbolic data, AND an operating system of immense complexity which regulates the production of each nanomachine and the regulation of each metabolic and structural component in patterns where groups of machine and processes unfold to produce a generative whole.
A blueprint may be used to build a factory. But in living systems, instead, there is a written instruction manual made of words, interpreted by a cooperative robotic system which interacts with software which "understands" how to translate those words into structures and further instructions,, with regulations in timing.
If the first group of builders builds the first room specified, copies his textts but highlights what the next builder should do and shouldn't do, and this process continues.
What should make this extraordinary, actually a miracle, is these processes have built in error correcting systems which can accommodate changing environments.
The individual words which are written for the builder to read may be "door" but the where and when of installation cannot be found in a simple sentence.
Great summary!
Great insight to understand life! But sill the question “where it is encoded?” is unanswered. I believe that species pass through generations not only physical characters but also experiences. I mean not only instincts, but also complex behaviour patterns (e.g. cuckoo chick ejects eggs out of the nest). The question is relevant not only to morphology, so what is the answer?
Kind of The Little Prince’s magic box answer: somewhere within the very first cell AND/OR somewhere outside the cell. The part “outside the cell” cannot be neglected, because “the law of physics,” at least at atom level (and sub-atom) provide infrastructure life is based on. Can we reject the idea of Cloud storage? Or, may be single cell encodes much more than we thought
We need to learn how to talk to cells, proteins to convince them to do what we want.
That's the idea!
This was fascinating.
Indeed it was! It was a privilege to host Michael at our conference.
Glad to have stumbled upon this channel, it is of spectacular quality.
Thank you so much for your interest! We are glad you are enjoying our content. Please visit our website for more information on the society behind this channel: semf.org.es/
4:50 - Missing the words 'as efficiently'. Or in other words - by better predicting the future
Interesting observation!
thank you algo for bringing me this amazing stuff!
We are very happy the algorithm is bringing this to the fore!
Please tell us about local growth factors for different facial features pls
I have seen this talk pop up again and again on youtube over a long time. Any news on this?
This was an update in the general context of the concept of Spatiality in our conference. Arguably a summary of previous, previously known work, but very relevant to this concept of space and shape.
Did he really make one of the best science lectures I’ve ever seen … And use Comic Sans font for his slide deck? 😂😂
Indeed that was the case! This may give a different status to the font from now on...
this is wonderful.
We are glad you enjoy our content!
I liked the talk, but rather than suggesting routes of future research, it was describing why the question is a question to begin with & that it is yet to be explained.
To be fair it does end by suggesting it’s unlikely purely emergent, but (based on the title) I was expecting some potential explanations of how it is encoded & more about ongoing research into those explanations.
That's a fair description. Michaels work is amazing and very promising but still early stages. We look forward to inviting him and his collaborators again!
Amazing!
so one starts out as a point in morphospace (I like that term) and gradually expands, stretches and shifts to occupy different regions. sort of like mthematical functions in a vector space? can we then work out the orthonormal equivalent in cell mrphology?
how many dimesnions can morphospace have as well?
Since very little of genomes is understood, how can someone claim that the information for body development is not contained in the genome?
Low IQ.same genome different anatomy he explains how.
how can the axolotl tell when the set point is reached? amazing
That is indeed part of the mystery!
What makes you think you can back-derive the genes from the anatomy? Certainly there no unique 1:1 function, at least in the backwards direction. Anatomy is a high-order effect of self-interacting generation, therefore there are many initial conditions that will produce the same result.
Where was such a proposal stated?
@@SEMF I thought that was the leading question for the 'anatomy compiler' part. He says you *should* be able to draw the anatomy and infer the stimuli for genes to become that anatomy. Perhaps later he says that is silly or something I'm not sure.
@@anywallsocket Not clear if that was mentioned in the end. Thanks for the comments!
Wow! My preconceptions are blown.
That's a great sign!
There must be some incredible interaction and communication between cells and their molecular signals.
That's right!
It is Memory, which is present all over your body, not only in the Brain, the rest is done by food intake.
If you eat a piece of bread it becomes you, if another person takes a bite from that bread it becomes his body with all features according to the memory collected.
That's a pretty accurate short account of it!
I recently wondered about this so thanks youtube. Being a c++ coder I was wondering where the memory of the huge design of our body is situated. I googled a bit and I found that we know at least some genes that code for some structure. Good to know that overall its still a mystery. Being in addition a mathematician I know that a fractal can be defined by a simple formula. Together with this feedback idea we can at least get some intuition.
We are glad to provide the content to support such thoughts!
Or some kind of recursive tensor with tightly distributed means.
@@kokopelli314 Could you expand what you mean by this? Sounds intriguing!
@@SEMF so fractals exist in state space where consecutive points have indeterminate or chaotic vectors. That is the N+1 point is indeterminate given variation of initial conditions, but the attractor, or pattern remains fairly consistent within a set of mean values. These values could be represented by tensor, or sets of n vectors with probabilistic values affected by a recursive function. Like you indicated in the video. It might look like magic, but a recursive function, acting through state space would be like the complex plane.
@@kokopelli314 Thanks very much for the clarification!
Splendid. Thank you..
Thanks for your comment!
DNA might describe a cell but may also convey rules to each cell like cellular automata does for each pixel. Maybe there's biochemical markers and cellular sense making of its environment.
That's a very real possibility, thanks for the insightful comment!
How then??! Tell us please!
I wonder if an individual cell can detect how many cells away is the nearest cell of a different tissue. like if a skin cell can detect how many skin cells are between it and the nearest subcutaneous fat cell, or if a liver cell can detect how many liver cells are between it and the nearest blood vessel cell, etc. Because if a cell *can* tell these things, then it seems to me like it would be a lot easier to figure out how anatomy emerges; each cell could roughly "triangulate" its position relative to other tissues in the body, and reproduce or die as needed to maintain an appropriate range of distances from the appropriate surrounding tissues. This is just a hunch but I'd be interested to hear what you think of it @SEMF
This is Faraday and Maxwell level science. Their explorations in electromagnetism led to electric motors, telegraphs, phones, computers, etc. What will Levin's explorations into the bioelectric role in morphogenesis bring?
We hope to be there to attest future results!
History reveals the weaponisation and marketisation of science by invested insiders seek prior advantage of insider knowledge, kept secret or surrounded by obfuscations.
The Field approach is NOT a control model but a mindset of control seeks according to its own conditioned thinking, under the mask of serving humanity, for possession and control. Hence The Biotech revolution of augmented or captive human to such systems is close at hand.
Living waters are structured by nanoscale quantum charge domains as a medium of translation of information unfolding from a whole. The pathological model rises from a sense of lack, exclusion, conflict, denial, set to defences that boost and reinforce the model as the basis for interpretation.
Wholeness is a balancing alignment of a life as contextual expression. But All the king’s horses and all the king’s men cant see it as that predicates on a broken life that must be fixed.
@@binra3788 Interesting considerations!
Future nobel prize winner.
for sure
Very likely indeed!
110%
Ok. Simple model problem. Suppose we want to generate a ball of cells with a certain number of cells as its radius or a certain distance as radius.
What would be a simple way to achieve that.
What would each cell have to measure, and what would be a mechanism to compare that to a number stored in for example the genome?
A ball seemed like the simplest Morpheus space I could imagine, maybe there are simpler once.
Simpler problems like these are probably important to investigate. Perhaps you would like to share your thoughts within our community: semf.org.es/participate/join.html
Perhaps DNA contains the blueprint and specs for making whatever needs to made by simply storing pointers. The pointers point to a specific blueprint stored in some huge universal database from which the cells somehow retrieve the required information. For example, Pointer 6582240352982549 references the geometry of my big toe. Questions remain: Where exactly is the database, and how is the data retrieved? If you can answer either of these, I will gladly share my Nobel Prize with you.
That's essentially what Michael's research seems to point towards.
Had a lot of thoughts about this topic. For sure, information component persist somewhere in encoded form
We are glad our highlight video got you thinking.
I ve been in medicine field since many years ago, and I can call it after more than 33 years of practice , that God is real. !!! Congratulations for your incredible videos!!!!❤❤❤
So if a salamander was born with a deformed leg, that gets cut off, does it grow back a deformed leg or does it fix itself given another chance to grow the leg.
From what we learnt from Michael, it would depend on what caused the leg to be "deformed" in the first place.
5:00 morphOspace
Awesome!!! I always wondered about how shape and symmetry were “encoded”. Since I was younger I noticed that people who have extra fingers usually are not symmetrical and if symmetry was encoded in the DNA a mutation causing an extra finger should produce it on both hands, at least there were two sets of genes to control ontogenesis of each body side the body, which does not make much sense.
Precisely!
This is explained at 12:30 in the following th-cam.com/video/S395qX6G6HM/w-d-xo.html
@@aj-uo3uh Thanks for the link!
He ends with the question. I am no further ahead than at the beginning.
This is the beginning of a scientific research field, it is natural to still be humble and stay sceptical.
14:56 - "GOALS in anatomical space" I hear echoes of teleology... and I'm going there :)
Some causally justified teleology is probably the way to go!
@@SEMF :) hehe - I wrote an essay, googleable, "Biological Teleology" as to my guess how cells can "choose what *will* be good for them."
@@anthonyrepetto3474 Interesting!
Morphology would have to be this way, because there’s too many unknown influences to build by sequence. “Shapes” will be encoded as modes (not static targets). I would guess it’s encoded on the cell level. Default mode would just be split, and other modes would be based on signals. To find the encoded “shapes”, you would have to locate all mechanisms that respond to signals of any kind, in a reasonably reliable way. Forms would be encoded as a byproduct of a massive mode mode hierarchy that determines each cells current role in the total system. If all cells are essentially the same, but specialize based on signals, to encode new forms, it’s all about the mode hierarchy. The mode hierarchy is going to be insanely huge, so identifying the levels and branches will be the main challenge. Once identified, individual sections of it can be reprogrammed with the appropriate signals applied to the appropriate levels. But preventing reversion would be a challenge when trying to alter a grown body.
That idea about modes is certainly plausible. Makes sense with the observed evidence.
@@SEMF Thank you for your videos, Micheal is one of the most brilliant people I’ve heard speak. Since the mode-like cell behavior is not crazy, here is my guess for where the forms are stored.
Assuming evolution is lazy/cheap, the easiest/cheapest way to encode physical functions (including “target form seekers”) would be to use the structure or “shape” of the control surfaces (the structures in cells that behave differently based on inputs; I don’t know the proper terms). This is explained at the end. Reading DNA is probably expensive, so you must store the real-time access information somewhere else. Putting all this together means you would expect DNA to encode instructions for an ultra general purpose biological machine that can copy itself, stay connected with, authenticate, and communicate with the copies, and change between a vast landscape machine types (resulting in a vaster landscape of possible behavior spaces), depending on its own state, signals from copies, and perhaps signals from non-copy “friendly” cells. You don’t design a body, you fine-tune a single super-cell that can copy itself and take on an infinite number of different roles/modes to fill its space in the environment. The cheapest way to store real-time morphology information (returning to the explanation) would be in the shape of the control surface, because its physical structure can be such that it reaches a natural balance for free. The analog thermostat example is appropriate. If the mercury has risen to the midpoint due to the temperature, you may ask “where is the form of 1/2 stored in the thermostat?” But of course it’s not, it’s encoded in the design of the control surface of the thermostat, which functions based on the environment it exists in.
DNA would encode instructions for building a cell-building framework (some kind of super-proto-cell), that would encode or re-encode a specific morphology in its control surfaces based on what type of cell it is, based on signals from other cells and it’s own state. Since the real-time form/morphology information is encoded in the physical structure, it would not appear like data, even though its structure is data driven. At no point would you find anything that represents a body part shape because it would actually be represented by the control surface shape that seeks the proper steady state, for its purpose, at its location, for the purpose of resulting in the desired overall morphological shape (even though it has no knowledge of this grand purpose). The actual forms we see when looking at bodies would be meaningless side effects from the cells perspective.
So, a person who is taller will have “taller” control surfaces. Not literally taller, but they will be constructed differently in a way that results in a steady state that results in a taller person. The control surface differences that encode real-time form information just by existing, could be revealed with machine learning by comparing them in detail (if detailed active-cell snapshots are possible) to the large scale morphologies, with enough properly tagged sample sets from one species.
Cells would have to somehow periodically do an expensive “check” of the DNA to make sure the control surfaces are still shaped correctly, otherwise bodies would drift off into strange shapes as they take on damage, and quickly stop working (because of the hierarchical/recursive nature of the signal-response chains). You should also expect some kind of checksum or redundancy somewhere in this process.
If true, altering DNA would be the most difficult way to manipulate forms, because you’re programming under several layers of abstraction. But it would be the most permanent. If you tried to manipulate the control surfaces, the form would revert after DNA checks. However, I suspect the surfaces are far more complex than this simplistic picture, and that there are infinite ways of manipulating the cell systems to capture new forms permanently by hacking the existing systems (coaxing the system into new steady states by pushing specific signals past expected boundaries; essentially causing overflow errors). Again, if true, every cell should know how to do *everything* that any other cell of the same species can do, if properly coaxed through the mode hierarchy.
@@909sickle Thanks for the very detailed and interesting comment!
The answer may lie in a new understanding of physics rather than biology.
We humans are natural-born engineers. As such, we model after machines not only isolated, naturally occurring systems, but also the basic laws of physics, sharing with machines a local-evolution-of-state `grammar'. However, there is a strong case to be made that it is this mechanistic paradigm which is to blame for the stagnation on many of the stubborn open problems in physics. The implications for biology are precisely what Michael is eluding to: `goal oriented' evolution in the sense that the goal is already `out there' - in the 4D block universe.
That's an interesting observation, how would you go about testing that model?
@@SEMF You can find a concrete proposal titled "Questioning the mechanistic-universe paradigm using chaotic systems" on my ResearchGate page (I would have included a link had TH-cam allowed). Still a preprint. Not an easy task finding a journal willing to publish such heresy...
@@yonatan2806 Thanks very much! Please feel free to join our SEMF online community and share your preprints and ideas here: semf.org.es/participate/join.html
Perhaps the nephrons forming the lumen are paying attention to the surface tension across their cytoskeletons. If multiple cells could bend together and achieve their desired surface tension amount together, then great. If only one giant polyploid cell is available, then that's fine too.
A very real possibility!
form, FIT, and function
The key is the distribution of regulatory substances in the body and their interaction with information/rules in the cell: if there is too much or too little of the substance, less or more will be produced. If you have rules for multiple substances, you get a stabile pattern (standing wave), like the stripes of a zebra, that define areas of the body. Think how this works on a scale with hundreds of regulators that create a complex pattern to define aspects of the morphology. If this system is perturbed, if will strive to get back into balance. If a new cell migrates into a tissue, its role is defined by the mix of regulators in its surroundings. The informational model for this would be a 3D pattern of Turing engines, interacting with each other locally, writing and reading streams of 'tape'. Each engine has hard-wired ROM data (DNA) as well as local RAM storage (RNA, methylizations). With the right kind of wiring and initial settings, the machines will interact to create an organism. If the settings are wrong, the organism will not converge and die off. Nature selects the working settings.
Very interesting thoughts! You may want to share them with our internal SEMF community: semf.org.es/participate/join.html
@@SEMF Thanks for the invitation, I feel honored and will grasp the opportunity.
@@SEMF Thanks, I feel honored by your invitation.
See also "Growing Neural Cellular Automata". (I would provide an actual link, but youtube doesn't seem to like comments with links.)
Thanks very much for the suggestion!
The answer to this question can be found by deeply understanding the mechanism of self-similar structure transformation (Fractals). Just as the lungs were mapped by Mandelbrot himself so too can anatomical structures (Mandelbrot BB. The Fractal Geometry of Nature. W. H. Freeman & Co. Ltd; 1982.). Just as the Mandelbrot set forms 3d acorns, so to do organs and anatomy form finite shapes and they dont "stop" they continue but they iterate on themselves until no more energy is left in the system in which they are derived from.
The key would be to use cross sections view (such as from fMRI etc) of development and with OpenCV to devise the baseline "fractal" that creates said shape. As to "where" this information is actually encoded, that answer becomes more clear when you start deconstructing all biology in terms of its baseline fractal. The encoding of this information is not present in the system directly, just as if we were to look at ChatGPT and view one of the transformations it wouldn't make sense as to how that singular transformation is able to perfectly structure sentences and the things it does. The information is encoded more in a gestalt way.
(Fractals of Brain, Fractals of Mind: In Search of a Symmetry Bond (Advances in Consciousness Research, No 7)
Thanks for the insightful comment!
good question
Michael outlines a very enticing answer!
The Cellular Positioning System is probably set by holographic interference transmission and reception by cell wall antennae.
the power of exaptation/re-purposing/comporting to novel contexts.
indeed!
So anatomy growth is a confluent rewrite system
Indeed! That's a very interesting angle! Many SEMF community members would love to discuss these ideas further. We invite you to join by following this link: semf.org.es/participate/join.html
There’s a game called cell lab where you make little multicellular organisms by defining a set of cell types, for each saying the angles of the descendants upon splitting, and what cell types the two descendants are. Combine this with the physics based precession of cells pushing and deforming each other and you arrive at some morphology. I’d say that encoding is not a great word for organisms that we know in real life, as code suggests it’s all layed out in readable form, when really it’s just a lot of evolutionary biases and constraints from physics, that is with the exception of the more organised aspects of dna, which aren’t the full picture.
That sounds super interesting! Why don't you share some links or info about that game with our community? You can join at no cost other than filling a quick form here: semf.org.es/participate/join.html
Would love to know what Rupert Sheldrake would say to this.
He has some pretty interesting theories about similar.
Congratulations ! It was not easy to summarize such a complex issue. I am not sure this riddle will ever be solved, not for technical reasons but idological ones . The denial of some kind of finalistic intelligent inherent to natural mechanisms make possible to go beyond mere description of facts.
I mean "impossible" obviously...
We are glad you enjoyed this video! Michael's summary was certainly well made.
So your saying we got the hardware but don't understand the software issues?
That's largely what we got from Michael during the Spatiality Conference.
@@SEMF right on bro.
I thought we already knew this. It’s not just life, but physics as well. Universal darwinism applies to all self-stabilizing systems ie any dissipative structure whatsoever - including physics itself. Clearly complexity and order supervenes on self-interaction at various levels of scale - it is not built into the initial conditions like your reversible, idealized, linear, physics would have you believe. Most phenomena are non-linear open dynamical systems, so if they are not autopoietic or at least self-stabilizing, they will dissipate due to the stochastic nature of reality, and therefore will not perpetuate long enough to converge onto self-replication and eventually become an ideal Quine.
First time I’ve ever wanted to hear the words “like for part 2”
We will invite Michael to future SEMF events! For now, you can watch the full talk here: th-cam.com/video/jLiHLDrOTW8/w-d-xo.html
Clearly. that anatomical information has to be in the genome, as the animal starts from a single cell. But that information is not in the DNA sequence. Sequence is a low-level language for making proteins. This language includes sequence of amino acids, but also meta information like starting and stopping points, where to snip, and regulatory areas.
There must be a higher level language layered on top of the DNA sequence. This language must be capable of describing anatomical structures by shape, size, and timing. What are the components of this language. What are it's equivalent of Words and Syntax? Or does it use some entirely different method of encoding information?
I seems completely invisible to us. Why? Is it so completely different from the way we think?
Very interesting thoughts! We encourage you to join our community to continue the conversation there: semf.org.es/participate/join.html
When we consider Evolution, how does that interplay here? I was always taught that evolution means mutations in genes which lead to a change in fitness. But if the morphological change is independent of genes, then that approach is wrong. So what exactly is evolution. Does that mean we have no clue how it works. If I'm not mistaken evolutions main supporting fact is that they compare the morphological tree based on fossils and the phylogenetic tree and make them fit.
Michael's work just points to spatial patterns of bioelectricity as other possible forms of information encoding, similar and complementary to genetic encoding.
For me the key has to be information. We know that DNA is the primary information repository.
Biology abhors inefficient processes, we know that only 1% of DNA codes for proteins. Therefore it seems obvious to me that the 99% of non coding DNA must be serving another function and is highly likely be the key to intra and inter cellular organisation through, as yet unknown mechanisms. It must be independent of neurological mechanisms or flora would be unable to regenerate form.
That's a great observation. Also non-genetic information, such as morphic function, can be relevant.
@@SEMFgood luck, I worked on cytokines. Even simple regenerative processes are horribly complex in higher eukaryotes and failed.
Best advice is to choose as simple a model as possible and the simplest process you can find. We are missing fundamentals, which is why I’d start with looking for additional information storage mechanisms which have to be in plain sight.
@@yp77738yp77739 That's a good point, what would be examples of such mechanisms?
@@SEMF We wouldn’t carry that huge “junk” load of conserved information. That 99% of “non coding” sequences remain largely unstudied.
My wild speculation as to how it’s read and the outputs would be unsupported by data If I were wanting to understand the mechanisms, the logical place to start are one or both of the gametes, all the necessary information must already be present prior to the zygote. The one exception could be if there were a maternal transfer of information whilst in the uterus, it would be easy to exclude this by looking at developmental issues in animals grown fully ecto uterine, which I believe can be now be done. Not my field of understanding though.
@@yp77738yp77739 Definitely doesn't seem like it is "junk", could be contextual information available for evolutionary exploration?
Perhaps thoughts are nothing but the experience of these feedback loops from a future ideal state to the present state, the present state incurring the form of the ideal future state in order to maintain or attain homeostasis towards the shape of the "body" the thought is pointing towards....on the higher level of self-organization of the human body, these are the higher dimensional possible/probablistic topologies of a resonant electric surface that the neurons of the brain create to mimic the internalized world picture it has in relation to these short lived ideal edifices of that they spontaneously create aka thoughts
That's a very real possibility here. Great comment!
This is epic! 👏🏾
Indeed it is!
Wonderful.
Thanks!