In my first post in this series, I wrote about the idea that reality is comprised of various “layers,” and I raised the question of how, in an evolutionary sense, do we “get” from one layer of reality to another. How did the molecular layer emerge from a universe that (at an early stage) included only atomic elements? How does the biological layer arise from the molecular/chemical layer?
We know many of the specifics of each “layer jump.” For example, we know that a type of chemical bond, the covalent bond (electron sharing) allows different elements to bind to each other and create molecules. In terms of how biological life started on Earth, we have some idea that it had to do with the evolution of self-replicating chains of nucleotides.
Is there a way to model these “layer jumps” in a general sense? If we could, we could make some incredibly interesting computer simulations. Perhaps we could model the emergence of biological life, of somatic forms, of social interaction networks, and eventually perhaps even intelligent entities. With sufficient processing power (maybe driven by quantum computing), we might be able to model an entire universe, including everything from the creation of galaxies and solar systems to the evolution of biological life to the development of culture.
What Is Evolving?
Broad models of reality are necessarily vague and imprecise; as one moves up the ladder of abstractness, one loses detail. It’s easy to visualize exactly what is meant by “cup of coffee,” slightly more difficult to get a concrete image for “caffeinated beverage,” and a word like “liquid” evokes no precise image.
A word to describe a “unit” of evolution must be abstract, and therefore vague. I’ve considered using the word “agent,” but many things that evolve (molecular structures, for example) don’t have agency. The word “node” might work, but it implies stasis. “Replicant” evokes the movie “Bladerunner.” I don’t have an ideal word. I’ll use “agent” for now. An “agent” could be a molecule (chemical level), a biological creature (somatic level), a poem (memetic level), or any “evolving thing” on any level of evolutionary space.
If the aim is to virtualize multi-level evolution (to simulate the “emergence” of biology from chemistry, for example), then an agent would be an object — an abstract entity with specific properties and attributes. The idea is to use a single class (or word, or variable type) to describe all types of evolutionary units so that our code (or algorithm) can generate new levels of evolutionary space without explicitly hard-coding them. The game Spore, for example, simulates multiple levels of evolutionary reality, but uses a different set of code and objects for each level (Spore is in no way an evolution simulator — in the game “evolution” is directed by the player — but I find the game idea to be interesting because it does in some way try to represent multiple levels of evolutionary space).
What’s An Agent?
In the model I’m contemplating, an agent has three basic properties:
- It has some degree of structural integrity.
- It has a genotype (a replicable code the determines potential structure) and a phenotype (a non-replicable structure or manifestation, the explicit and instance-specific expression of the genotype).
- It has a way to interact (exchange information or “communicate” with other agents of its type).
In the realm of biological (Darwinian) evolution, the genotype is DNA, the phenotype is the animal (or plant) body. The DNA genotype is replicable (via sexual or asexual reproduction) while the phenotype (the body) is not.
In the realm of molecular evolution, the “genotype” is the molecular structure (H20, for example), while the “phenotype” is the actual material (water, ice, steam). Molecules “interact” with each other via chemical reactions; they form and break chemical bonds. A water molecule can be broken down into its atomic elements (oxygen and hydrogen) but not without energy input; the water molecule has some degree of structural integrity.
On the memetic (idea, symbolic, language) level, the genotype is the medium while the phenotype is the case-specific expression. The medium could be a human brain, symbols written on paper, or a hard disk. Replication could be via hand copying, “word of mouth” (speaking and listening), or digital copying. Ideas “interact” with other ideas via human thinking and speech (and creating and viewing media).
But this is all just analogy, right? Biological evolution is the only “real” type of evolution, isn’t it?
Maybe, but I’m not done with my analogy. And if there is a way to represent each element of the model in code and run it as a simulation, maybe the model can be more than an analogy (not a theory exactly, but at least an operating simulation).
What Is Evolutionary Space?
When agents interact, they create a “space” in which evolution can occur. Biological evolution occurs within a somatic space; bodies interacting. Animals fight, seduce, mate, and do or do not pass on their genes (thus replicating their aspects of their genotype). The “space” can be visualized in terms of a network of interacting nodes. Each body (animal or plant) is a node in the biological network.
On a planetary level, molecular evolution can be said to occur within a geological space. Materials (rock, water, lava) “interact” geologically, the result is more or less of various molecular compounds, as well as the creation of novel compounds.
As Richard Dawkins suggests, ideas also “evolve” in their own evolutionary space. Dawkins proposes that “minds” are the home of memes, but I would expand the cultural space to include all media, of which human brains are a part. We externalize and formalize our thoughts by writing them down (or typing them up). Books, film, TV, written and recorded music — ultimately they are all forms of crystallized thought and expression.
Each “evolutionary space” has its distinct types of agents and interactions. The “higher” spaces are subsets of the “lower” spaces (for example, all biological interactions are also chemical, but not all chemical reactions are biological; biological evolutionary space is a subset of chemical evolutionary space). Rules that describe interactions in a lower space are “true” for all spaces above, but don’t meaningfully describe the interactions that occur within that space. If two people go out on a date (a social-emotional type interaction) then they will not violate any laws of physics. However describing what happens on the date in terms of physics will probably not give us any meaningful information about what happened on the date.
But Is It Evolution?
Along the lines of the argument I just made, you could say that trying to describe all levels of reality in terms of evolution is a lost cause — you’re never going to get any meaningful insight about culture or human thought by using the analogy of Darwinian evolution. I agree to some extent. However, what I’m interested in is laying the groundwork for creating a simulator that can generate emergent levels of reality (without explicitly hardcoding them), and to do that I think it is necessary to pursue a “general theory” of evolution that extends beyond biological evolution. With the understanding that this might be both impossible and/or useless, I’ll proceed anyway.
In Darwinian evolution, the basic elements are:
1) Heritable traits (via DNA), and natural variation in traits and combinations of traits (via DNA mutation and sexual recombination).
2) Natural selection or “survival of the fittest”; organisms that survive to a reproductive age and manage to reproduce manage to pass on their genes (and thus their traits). Fitness traits are not universal, but depend on “what works” for the particular environment. For example, physical strength may be helpful if food is abundant, but small body size may be a more important fitness trait if food is scarce.
3) Genetic drift — random changes in the frequency of genetic variants, especially those that do not strongly influence survival or reproductive success.
Do these same basic elements exist in evolutionary spaces besides the somatic/biological space? To some extent they do. For example, a particular molecular structure can be replicated via a number of different chemical processes. The gemstone malachite — Cu2CO3(OH)2 – is formed when copper ore is exposed to the elements. It is usually found near limestone, which supplies the carbonate. To say the chemical structure of malachite is “inherited” would be pushing the analogy too far, but it is certainly replicated (via crystallization).
What about fitness? Fitness traits will be different for each agent type, within each distinct evolutionary space. In the most general sense a fitness trait is whatever allows an agent to persist and replicate. In the geological space, where molecular agents (compounds) evolve, we could define fitness as:
1) Having structural strength — not being prone to easily dissolving.
2) Being formed from abundantly available elements (like oxygen, hydrogen, and carbon).
3) Not requiring extreme conditions to be formed (extremes of heat and pressure).
The list could go on. Fitness traits in the somatic space (biological agents) will look more familiar:
1) Having a robust immune system to fight off pathogens.
2) Having defense mechanisms (tough hide, fast speed, thorns) to fight off predators.
Fitness traits are never universal — they always depend on local conditions. In environments where conditions vary dramatically, adaptability is a premium trait. If the environment is more stable, niche specializations can be highly effective. And there are always trade-offs. Panda bears, for example, never run out of food as long as they are living in bamboo forests. But take away the bamboo, and panda bears are hosed.
I’ve written about the relative fitness of memes here (scroll down to the lolcat picture). In short, memes that are easily replicable (say, via link forwarding), catchy, short, easy to understand, and have small fuzzy animals or young beautiful women as their subject matter will be “more fit” than obscure, unsexy, difficult-to-understand memes (like this blog post). Memes that are useful or interesting to a large number of people will also fare better (I can see from my own traffic stats that posts on video games and health get more interest than those about evolution analogies).
How Is A New Evolutionary Space Created?
Is it possible to describe in a general (yet still meaningful) way how one evolutionary space, or level, emerges from another?
If an evolutionary space is functionally a network of interacting agents, then to create a new space, a new type of information exchange is needed. We could use the phrase “mutant node” or “transcendent replicant” to describe an agent that exists within one evolutionary space, but via a novel transaction mode or information exchange creates a higher level space.
One example would be the emergence of animals with specialized cells (tissues). Before cell specialization, unicellular/protozoan and simple multicellular animals interacted and evolved in the biological space. These simple animals competed (and still compete) for food, light, and other resources as all biological entitities do. On this level, actions and interactions are primarily metabolic (cellular digestion and elimination). As cell specialization emerged, animals began to take on distinct forms. Loosely organized clumps of cells evolved into bodies, with different tissues performing different functions. This opened up a new evolutionary space, the somatic space. Different animal and plant bodies created a new network, a network of somatic interactions (fighting and mating, for example). Different fitness traits (strength, speed, being poisonous) could start to affect the course of evolution in this new space. In short, animals who evolved specialized cell functions were the “mutant nodes” that opened up a new evolutionary space (the somatic space, made of of somatic interactions).
Going “down the ladder,” how did the biological space emerge from the geological space? There are many theories about the origins of life, but they all involve nucleic acids. Of all possible molecular compounds, protein molecules, specifically chains of nucleic acids, are the mutant agent that creates a new network. The interaction and production mode (how new agents are replicated) changes from chemical reactions to protein exchange and assembly. Of all possible molecules, nucleic acids were the “mutant nodes” that allowed the biological evolutionary space to arise out of the geological evolutionary space. A new class of interactions was created (biological interactions).
Back up the ladder, what new space emerges from the somatic space? I think the emergence of animals that are able to experience, express and interpret emotions opens up a new social-behavioral space. Suddenly, social bonds are possible, and traits like loyalty, empathy, and the ability to manipulate and deceive others start to figure into survival and reproductive fitness. Of course this space can still be described by Darwinian evolution, but instead of the stag with the biggest antlers winning the right to reproduce, it might be the most charming or wily or even sensitive monkey who gets the honors (in primates the “alpha” isn’t the always the biggest or strongest — it’s often the most socially connected and empathetic). The development of a sophisticated nervous system (that facilitates emotion and memory, in addition to perception and instinct) creates a new network of social-emotional interactions (friendship, courtship, and long-standing rivalries).
In some primates, the brain continued to grow in size and power. At least one hominid species (us, but possibly Neanderthals as well) developed the use of sophisticated symbolic language and the ability to visualize scenarios (directed imagination). Along with the new brain power came novel behaviors like tool-trading and specialization of labor. What emerges is a new cultural-technological space (memetic evolution). What happens in this space? All of human history, for one thing, but on other planets there are (or were, or will be) likely parallel evolutionary spaces.
On the “higher” levels, entities are made up of multiple agents operating on multiple levels. A human being isn’t really a single agent. A human being is a physical body, and that body may or may not reproduce sexually (thus producing another human body), but that isn’t the same as replicating a human being. What’s actually getting copied are some of each parent’s genes (DNA), and the general somatic form (human body). In addition, certain memes (ideas, cultural values, family traditions, words and symbols) are passed down culturally (via speech, writing, and other forms of media).
It’s possible that some future technology will allow human beings to be replicated as agents. Imagine an extremely high resolution 3-D biological printer that made a copy of your body down to the last axon connection. Unless there is a unique disembodied human “soul” (I don’t think there is) then that entity, once created, might experience itself just as you experience your self. That would be true replication. Of course, the entities would instantly diverge experientially, and become different people. Simply the knowledge that you were “the copy” would profoundly effect your experience, even if you were identical down to the last quantum particle. The film Moon provides a brilliant dramatic example of how such a situation might play out.
The Big and the Small
There is a micro/macro division within each evolutionary space. Elements, atomic agents (micro), evolve within an astronomical space (gaseous clouds, stars, galaxies — the macro side). Molecular evolutionary agents (micro) evolve within the macro planetary/geological evolutionary space.
Is a planet (macro side) a distinct type of agent? I don’t think a planet can be said to have a genotype (even by the loosest analogy), nor can it be replicated. So no, a planet isn’t an agent. Still, there are interactions that occur on the macro side that influence the evolutionary paths of agents on the micro side. If a planet crashes into its sun, all compounds (molecular agents) are destroyed. The process of a planet crashing into a sun can’t be described via the processes that govern the evolution of micro-side agents (chemical reactions, covalent bonding, crystallization, etc.). Gravitational forces and orbital geometries are what determine if a planet is swallowed by a sun (or not). So the macro side is important, and operates by distinct rule-sets, even it doesn’t contain its own evolving agents, per se.
Biological agents evolve within ecological space. Somatic agents (animals and plants with cell-specialized bodies) evolve within territorial ecological space (even plants behave territorially — competing for sunlight — even if they don’t have any territorial feelings). Emotional agents (animals with a capacity for emotional expression and perception) evolve in a social space (kinship groups).
The Emergence of Interiority
At some point in a local evolutionary timeline (like the evolution of life on Earth), consciousness may or may not evolve. Evolution does not evolve towards consciousness (or towards anything, except fitness — and that’s a moving target). However, having a powerful brain can be a powerful (if costly) evolutionary advantage in many ways, so it may be that this niche gets filled under many different local conditions (we won’t know until we have some more data concerning life on other planets).
What’s the Point?
What’s the point of developing a broad evolutionary model (or, if you prefer, evolutionary analogy)? Personally I think it’s an interesting thing to think about and attempt to understand. In addition, it might help lay the groundwork for creating an interesting general evolution simulator (a “Create Your Own Universe” software app).
You would need a massively powerful, parallel, quantum computing system to create any kind of interesting simulator. It would have to be parallel, processing many transactions at once, because that’s the way the world seems to work (events don’t occur in an orderly linear sequence — they occur all at once in a mad jumble). It would have to be quantum because that’s also the way the world works — as far as we can tell quantum mechanics is the lowest discernable operating system of the universe, and an accurate simulator would probably have to take advantage of certain properties of quantum computers (like being able to generate truly random numbers).
In addition to the right kind of hardware and operating system, you would need the right algorithms.
Years ago I created my own simple biological evolution simulator. Each “gene” in the software was a function that attempted to perform certain actions in the environment, or conferred certain traits to the agent’s body. It was a fun project, and various interesting behaviors emerged within the ecologies of virtual creatures. Of course, because I “hardcoded” a single evolutionary level, there was no way anything truly novel could emerge from the simulator, even if I threw unlimited computing power at it.
I don’t know how to program a multi-level evolution simulator. I do think such a simulator would include:
1) A flexible “holonic” agent variable, that could relate to other agent variables (of various levels) in flexible ways.
2) Emergent network types (ecologies) defined by how various agents and groups of agents exchanged classes of information.
3) A general (and generational) system of agent replication and mutation.
Maybe None Of This Is Necessary
It could be that the only reason to explore how evolutionary levels emerge from one another is so that we can understand how the process works. In order to simulate the complexities of a virtual evolving universe, it might be possible to plug in the “lowest level” rules (maybe some aspect of string theory, or whatever Theory of Everything happens to unify quantum dynamics and gravity), and run the simulation on some kind of mega powerful quantum computer. Distinct evolutionary spaces might evolve on their own, without a special algorithm that describes the “level jump” process.
Computer simulations don’t need to understand what they’re doing, and neither do the humans who create and operate the simulations. We may eventually end up inventing a black box evolution machine. An algorithm can be put to work even if it is so complex and bizarre as to defy human understanding.
We like easy, elegant, maximally reduced solutions, that facilitate an intuitive understanding of what is happening. For example, Newtonian physics is the best method for human beings to understand concepts like mass, force, and acceleration. But that doesn’t mean that Newtonian equations are required to simulate the way objects interact in space. Einsteinian or quantum physics could work just as well (even if the equations are messier and make comprehending what is happening more difficult for human beings).
Even if a special “level jump” algorithm turns out to be unnecessary, understanding how evolutionary “levels” or “layers” emerge is still important. If there are consistent patterns to be found, understanding those patterns could help us understand how complex chemistry, and life itself, arises on other planets. In addition, we could use the “level jump” algorithm to create interesting evolution simulators before we have access to a) a workable T.O.E. and b) computers powerful enough to simulate extra-biological evolution from the “bottom up.”
Ideas expressed in this post draw heavily from the writings of Ken Wilber, Richard Dawkins, Susan Blackmore, Daniel Dennett, Stuart Kauffmann, and of course Charles Darwin. The only idea that is new (if it is new) is the idea that a new evolutionary space is created by the emergence of a new type of information exchange (and thus a new network type).