In my last question I asked why we don't see increased complexity in artificial life simulations of evolution. It seems I had fallen for a common misconception, that evolution was about improvement by increasing complexity. One comment discussing that post read

"... he [David Deutsch] is falling for one of the biggest misconceptions about evolution that you can, that evolution is about improvement. Evolution has simply only ever been about change..."

However, when you look at the history of life you see increases in complexity. You see this increasing complexity evolving over billions of years, suggesting that it requires an explanation.

My question
If evolution is not about increasing complexity then how does so much complexity evolve?

  • 3
    $\begingroup$ The word improve can mean many things. There are many ways to "improve" - developing intelligence, flying, surviving extreme temperatures, good eyesight, good sense of smell, fast running, echolocation, etc. We just happened to develop some of those. I think it's better to call it complexity than improvement. $\endgroup$ – martinkunev Jan 7 '16 at 13:27
  • 2
    $\begingroup$ Comments are not for extended discussion; this conversation has been moved to chat. $\endgroup$ – WYSIWYG Jan 8 '16 at 8:27
  • 1
    $\begingroup$ @KyleStrand The question had just too many long comments which nobody would care to read. Moreover, as I can discern, comments and answers are getting repetitive. The usual SE practice is that when the OP addresses the point raised in the comment, you delete the comment. We are not hostile towards this question; unfortunately this topic is such that everyone wants to give their opinion whether or not they have a proper understanding of the subject. Moreover when a question hits HNQ, you have to be really careful. If we were really hostile then we would have closed the question. $\endgroup$ – WYSIWYG Jan 9 '16 at 6:28
  • 1
    $\begingroup$ You need to clarify what you mean with 'improvement', as the answers show. This is a vague term, and it is unclear if you are referring to improvement as in higher fitness, higher complexity (which is in itself also rather vague) or something else. $\endgroup$ – fileunderwater Jan 12 '16 at 9:18
  • $\begingroup$ Change happens for the "benefit" or "detriment" of an individual. If it is detrimental, that individual stands less of a chance of contributing to the species' "gene-pool". Therefore beneficial changes are overrepresented in "gene-pools". Am I missing something here? $\endgroup$ – James Jan 18 '16 at 7:43

I think possibly the problem here is the way you're approaching the issue.

You're considering improvement as anything that increases the abilities or complexity of the organism—that isn't necessarily what an improvement is though. The outcome of natural selection is that the organism best equipped to survive/reproduce in a certain environment is the most successful. So, for example, thermophillic archaea do much better in 60°C-plus pools of water than humans do. Our capacity to process information, use tools, etc. doesn't actually confer much advantage in that situation. And there can be downsides to that kind of complexity as well, requiring more energy and longer developmental periods. So, natural selection in 60°C-plus pools of water gives you archaea, and in (presumably) the plains of East Africa, it gives you humans.

The comment you quote mentions sickle-cell anaemia, which is a different example. While there is little benefit to having the sickle-cell anaemia allele in a temperate region, in those regions where malaria is endemic, heterozygosity can provide a survival advantage, and so the allele is maintained in the population. If you're someone living in a malaria-endemic region, and you don't have access to antimalarials, heterozygosity for the sickle-cell anaemia allele is arguably an improvement. It depends entirely on how you define the word.

The fundamental principal of natural selection is that it favours the organism most suited to a particular environment. But, that isn't always the most complex organism. It's important not to confuse human-like with better. It isn't the universal endpoint of evolution to produce an organism similar to us, just the organism most suited to the environment in question.

Also, to briefly address the previous question you asked—you asserted that we must be missing something from the process of evolution because we were unable to simulate it. You also pointed out that (in your opinion) we have sufficient computing power to simulate the kinds of organisms you're referring to. But natural selection is intrinsically linked to the environment it occurs in, so the simulation wouldn't just have to accurately simulate the biological processes of the organism, but also all of the external pressures the organism faces. I'd imagine that, in simulating evolution, that would be the real obstacle.

  • 2
    $\begingroup$ Comments are not for extended discussion; this conversation has been moved to chat. $\endgroup$ – WYSIWYG Jan 8 '16 at 8:34

Evolution is simply a process of change. It is a change in trait values of populations over time. It results from four mechanisms: mutation, migration, drift, and selection. The first three lead to random change from one generation to the next, which may increase or decrease fitness, while selection will generally lead to adaptation (relatively increased fitness in subsequent generations).

"Evolution means change, change in the form and behaviour of organisms between generations. ... When members of a population breed and produce the next generation we can imagine a lineage of populations, made up of a series of populations through time. Each population is ancestral to the descendant population in the next generation: a lineage is an ancestor-descendent series of populations. Evolution is then change between generations within a population lineage." - Ridley, Evolution, Page 4.

This is what Darwin termed "descent with modification". Later in Ridley's book he goes on to say something which is important to for evolutionary biology; why is there so much adaptation?

".. not every detail of an organism's form and behaviour is necessarily adaptive. Adaptations are, however, so common that they have to be explained. Darwin regarded adaptation as the key problem that any theory of evolution had to solve. In Darwin's theory - as in modern evolutionary biology - the problem is solved by natural selection." - Ridley

Another good clue as to what evolution really is comes from the Charlesworth & Charlesworth book:

"Evolution means cumulative change over time in the characteristics of a population of living organisms. ... All evolutionary changes require initially rare genetic variants to spread among the members of a population, rising to high frequency..." Charlesworth & Charlesworth, Elements of Evolutionary Genetics, page XXV

Basically the random mechanisms of evolution (mutation, migration, drift) are not as good at making rare beneficial alleles spread through a population as selection is. Selection is the major mechanism that should, as a general rule, fix beneficial alleles in a population. Drift, mutation, and migration will rarely cause the beneficial (adaptive) alleles to fix. Furthermore, mutation will generally have deleterious (maladaptive) effects according to Fisher's geometric model of adaptation.

You can read more about the process of adaptation and why selection doesn't guarantee adaptive evolution in my answer here. Briefly, selection will lead to adaptation if there is sufficient genetic variance in fitness, selection is a constant from one generation to the next, and genetic correlations do not impede the response to selection. Furthermore, the other evolutionary mechanisms can counteract selection, preventing adaptation. These are some of the reasons that simulating evolution accurately is so difficult.

The reason that we can't say complexity increases by evolution is that none of these mechanisms give a consistent increase in complexity. While mutation, migration, and drift will have random effects on organismal complexity, fitness (thus selection) may have some relation to complexity. To evolve, some degree of complexity is required such that the minimum conditions for evolution can be met. However, selection should favour the most fit genes over time, which depends on the niche/adaptive landscape and genetic variation available. Selection in the real world (as opposed to alife* world) would, as an approximate rule of thumb, favour an intermediate level of complexity where fitness is optimised (individuals are good at producing offspring in their niche) with minimal wasteful complexity (complex structures that do not increase fitness).

In summary, to answer your question, we see so much improvement because of selection, which leads to the process of adaptation, but adaptation does not equate to increasing complexity. The key to understanding your problem is an understanding of the difference between the process of evolution (change) and the process of adaptation (improvement), and the difference between optimality and complexity. In the world of alife simulation complexity $\equiv$ adaptedness, in the real word complexity $\neq$ adaptedness.

Good reading can be found in a link that AMR posted in a comment to another answer.

* Artificial life (alife) simulations of evolution generally use complexity as a proxy for fitness such that selection will be directional for increased complexity

Just as a response to a comment you made under your question, as to why simulations don't produce "stylized facts found in real evolution": Scientists understand quite well how evolution works (as explained in my answer, is a result of selection, genetic (co)variation, and population demographics), however, simulation to produce "stylized facts found in real evolution" would require a complete and precise history of the selection, genetic (co)variation, and population demographics that have existed since the origins of life. That is why simulation does not work like you think it should.

  • 1
    $\begingroup$ Concerning your edit: No, no, I am not talking about an exact reproduction of how evolution worked out on earth but about stylized facts, that is e.g. emergence of real innovations. And it is not about me but the whole Alife community - please see also the links given in the new edits here: biology.stackexchange.com/questions/42033/… $\endgroup$ – vonjd Jan 7 '16 at 19:26
  • 2
    $\begingroup$ @AMR... by the way, the quote continues: "the only time a fact was mentioned was when I mentioned it, and that was considered to be in rather bad taste." - so he was the smartest guy in the room surrounded by idiots... by his own account... why am I not impressed? $\endgroup$ – vonjd Jan 8 '16 at 8:30
  • 1
    $\begingroup$ @vonjd John Maynard Smith was often the smartest guy in the room... $\endgroup$ – rg255 Jan 8 '16 at 10:49
  • 1
    $\begingroup$ I am now reading through your answer again. So you are saying the evolution and adaption are two separate things and adaption is the process by which improvement comes into the world? $\endgroup$ – vonjd Jan 8 '16 at 15:46
  • 1
    $\begingroup$ Ok, so that would mean that we are missing something in this sub-process of adaption. That there are some details that we have not understood well enough yet to be able to simulate them. $\endgroup$ – vonjd Jan 8 '16 at 16:15

It might help to not think about evolution as a process at all - it tends to imply some sort of planning or goals or something like that. That's not what evolution is - evolution is simply a fact. When we talk about "the evolution of humans", we're describing the history of various human precursors. Evolution is basically a historical record of things that worked in the past in a given environment.

Most people tend to antropomorphize evolution, give it goals. There's no such thing, and it just makes you even more confused. There's nothing paradoxical about "evolving to extinction" - evolution is not a path from a base organism to an improved organism. It's simply a history of the changes that survived and thrived in a population. Sometimes that's because those changes gave the individuals and populations a better chance of surviving in their environment, so those traits became more and more prevalent in a population - for example, skin turning to hardened skin, turning to armour plates or weapons, or a better beak allowing it to reach into a food source that isn't available to others. Sometimes, it's simply dumb luck - don't forget that there was a point where the whole (pre-)human population was reduced to a ridiculously low number (I think it was something like 10 000 individuals or maybe even less). It would only take one local catastrophe to kill off the whole human species, no matter how "improved" and "advanced" we might consider ourselves to be.

Another rather brutal example would be the evolution of photosynthesis - when the atmosphere started filling up with free oxygen, it killed off almost all life on the Earth. Sounds like an improvement? Getting rid of your competiton? Well, it also fueled a massive growth of new species that were not only adapted to an oxygen atmosphere, they used it as a source of energy! Not only would they thrive on the "waste products" of the photosynthesers, they even consumed them.

Even if you wanted to describe evolution as a process that improves fitness, you must not forget that a change that improves your reproduction rates in one kind of environment can hinder (or kill) you in another.

When pre-Koala bears drifted to be exclusive Eucalyptus-vores, it gave them an advantage - they had a food source noöne else can use. But it also made them 100% dependent on Eucalyptus. When Eucalyptus dies, they will as well. Something that was arguably an improvement can easily be the thing that kills of your entire species. It only "improved" their ability to survive and thrive in one specific environment - it also entirely locked them in their niche.

In summary:

  • Evolution doesn't have goals, so it's weird to say "evolution is about improvement". Random changes have a tiny chance of becoming (locally) useful traits, and useful traits have a tiny chance of becoming entrenched in the population, and thus forming a new species over time. It's a history of changes, not a prediction of the future. The great thing about Darwin's Theory of Evolution is that it predicts what kinds of changes are possible (and which are impossible!) - for example, that complex systems can't arise out of the blue, or that different branches of history ("evolutionary tree") cannot exchange traits.
  • Almost all changes also have their drawbacks - it's a balancing act. There's some great examples of changes that are almost universally good - sexual reproduction and human-level intelligence are a great example of something that works in almost any environment. But even so, there's still examples of where they didn't "win" yet. There's still asexual reproduction on Earth, and most of Earthly life doesn't have human-level intelligence yet. Lions do not rule the world, even though they're apex predators in some environments.
  • 2
    $\begingroup$ @vonjd It happens all the time. The Koalas are one example. Cave fish evolved from normal fish, but they lost their pigmentation and sight - "became less complex". Species change their diets all the time (e.g. omnivore -> herbivore -> carnivore -> ...). That's why "devolve" doesn't make any sense - there's simply changes. Are there examples of a multi-cellular organism becoming able to work as a single-cell organism again? In modern history, probably not (though you might count "colony" cells). What do you really mean by "complexity"? Are rats more complex than fish? $\endgroup$ – Luaan Jan 6 '16 at 15:21
  • 5
    $\begingroup$ @vonjd If you look at genetics, for example, humans are one example - on of our chromosomes is damaged, and we "lost" others. All Mammals have two-colour vision by default, because some common mamallian precursor lost two colours - once you go far away, tetrachromancy was the standard. Some mammals got one colour "back" (including humans). We no longer have a tail, and our nails are almost nonexistent. Our intestines got comparatively shorter over time. Does this mean we're "less complex" than primates or other mammals? Is a plant with 100 chromosomes more complex? $\endgroup$ – Luaan Jan 6 '16 at 15:24
  • 5
    $\begingroup$ @vonjd I just gave you one - cave fish (and other cave dwellers) lost sight. The thing is, "in the capability to process information and the sophistication of the internal representation of the world" is a gigantic topic we can't really quantify well. If we could, we'd already have (or be on the road to) "real" artificial intelligence - we don't know what makes intelligence "tick", even, say, dog-level intelligence. We know tiny subsets (mostly the kinds where there's a direct A -> B relationship, like reflex movement). $\endgroup$ – Luaan Jan 6 '16 at 15:30
  • 3
    $\begingroup$ @vonjd Because most of evolution simulations have very big fitness penalties for code size. This is mostly due to two things - one, the larger the code and complexity, the slower the simulation, and two, you usually use evolutionary algorithms to try to find a cheap solution to a problem. If you remove the complexity penalty, complexity is going to explode all over the place. In nature, there are still costs to complexity (it takes energy, resources, ...), so it isn't unbounded either, but it's a lot less strict, since it's usually relatively small compared to e.g. the whole body. $\endgroup$ – Luaan Jan 6 '16 at 15:52
  • 4
    $\begingroup$ @vonjd I'm not saying waiting for more computing power and storage size is a good strategy. We need to figure out how intelligence works and find a way to reproduce it - I don't think we necessarily need more computing power and storage than we already have to run an AI; it just isn't enough to semi-randomly create it. But yes, if you had unbounded storage space and computing power, you'd get to intelligence eventually. The universe is fundamentally very simple, and it produced intelligence anyway - it just took a huge amount of work, and lot of time. $\endgroup$ – Luaan Jan 6 '16 at 16:34

I'm going to chime in here. As both a scientist and a software engineer.

Firstly, evolution is not about improvement at all. It is about survival and random change. There are as many if not more mutations that are disadvantageous. But they tend not to survive.

On the other hand, genetic algorithms are an attempt to use a similar process of mutation and survival of the fittest.

But the first step in a genetic algorithm is to define a fitness function. This function will cull the weakest algorithms, just like an environment kills life in the real world.

A good primer on Genetic Algo can be found on https://www.youtube.com/watch?v=qv6UVOQ0F44

However that fitness function will only optimise for certain goals. For example, a badly tuned fitness function will end life on earth either by the paperclip apocalypse, or giving rise to skynet.

In these cases the algo is not improving towards the goals you want. But never the less it improves.

Another complexity is that, genetics is a very greedy optimisation strategy. Mutations tend to be small, because large mutations tend to more often move away from optimal solutions. This means that evolution can only find local maximas and will often miss the global maxima.

Hence improvements can only occur when there is a small tunneling cost to the new maxima.

An example of this can be found in mammalian eyes. Our optic nerve passes through retina and connects to the front of our retina, and physically blocking the retina from doing an optimal job. If evolution were able to find a global maxima, then mammals would have been able to evolve to have squid like eyes, which route from behind.

Moreover, had evolution been about pure improvement, then we should have evolved away our blind spot many many generations ago.

However, human ancestors have rarely been attacked by circles and crosses that are precisely spaced apart in the African continent.

Saying that evolution is about improvements is like setting up a school where there is no teaching, and every year you expell the bottom 10% of the students.

  • 2
    $\begingroup$ If your answer contains no scientific claims/statements how can it possibly serve as an answer on a biology Q&A site? $\endgroup$ – rg255 Jan 9 '16 at 9:03
  • 1
    $\begingroup$ @rg255 Logic. The first half of the post merely reiterates definitions. The later half saree the consequences. I'd admit that I have no basis for the statement of no quantum leap mutations a la x men. $\endgroup$ – Aron Jan 9 '16 at 9:07
  • 1
    $\begingroup$ You cannot compare GA with actual process of evolution because in reality the fitness function keeps changing. In reality fitness cannot be assessed unless selection happens. $\endgroup$ – WYSIWYG Jan 12 '16 at 9:00
  • $\begingroup$ Sorry, You are saying that inter-species competition does not cause for simpler organisms to become prey and be vulnerable to extinction. Evolution caused complex functions of motility and defense to be added to every species on earth, even plants all have biochemical and physical defense, and simpler organisms that exist in extreme environments use the environment as a defense, without which their simplicity would cause them to be decimated. $\endgroup$ – DeltaEnfieldWaid Jul 21 '17 at 22:36
  • 1
    $\begingroup$ @comprehensible Correlation is not causation. Complexity is only one of many solutions to the ever changing problem. $\endgroup$ – Aron Jul 23 '17 at 12:22

Evolution produces branching trees, and branching is multiplicity.

Fitness is associated with complexity, and, with radiation into more or less challenging environments. Fitness increases with versatility and added functions, in equilibrium within one niche, and for change across niches. Are cold blooded animals simpler than warm blooded ones? the consensus is that they are simpler, less apt and globally superseded. Even if a lizard has as many genes as a human (40,000) it is less complex than a human.

Take Motility for example. Locomotion is complex, compared to passive or sedentary displacement. Most prokariotes have evolved some kind of motility, cilia and flagella. The simpler procariotes were eaten out of existence, as the motile ones grew to dominate and pervade. There is a predator prey issue which has resulted in the extinction of simpler, slower species and in the promotion of more complex ones. The ones that survived did so by adding defense functions.

Life is thought to have started in simpler environments with less biochemical and physical fluctuations than it later evolved into. Eukaryotes have not evolved back into prokaryotes, even though they could, and eukaryotes have more scope for complexity, same as lego blocks in multiple numbers are not as simple as single blocks.

Evolution is also about the blind use of an initially small but potentially much larger memory bank of many gigabytes."Gene Duplication is believed to play a major role in Evolution."

Unless life began in greater quantity than it now exists, evolution requires that natural processes have, over time, increased the total quantity of genetic material (DNA) present on our planet.

I'm going out on a pirate's plank of logic here. sorry about that.

Survival fitness is about increased complexity when the environment is increasingly complex. Evolution Causes complexity to occur...

Change is an additive process, and the more change is provoked, the more added functions tend to result along.

The more complex the path has been to arrive at the current stage of the species, the additive complexity rises. (Also the DNA keeps on record the genes from old environments to not lose many precious years spent finding useful genes/biochemistries, while adding new ones). However evolution can be about the conquest of less complex environments:

Put a fish into a cave with no lights, constant temperature, and simple tasks, it will lose some of it's complex genes and may, over time become genetically simpler, than a fish living in a river. It requires less senses, less thermal adaptations, less locomotion pressure and less species competition. It is rare for species to retrograde generally, they tend to extend their range, but in deep sea and caves, locomotive and biochemical retrograde can happen.

Increased size gives increased fitness in most settings: larger metabolic reserves, less sensitivity to change, bigger predator pray advantage... and bigger size means more cells, different locomotion pressure, which means different metabilic resource distribution(lymph, digestion, blood), and That is that's the essence of your complex topic: Do environments encourage complexity? if so, how "complex" are environments in the universes land/sea biomes? Geology, climate and hydrology are of incredible complexity... So... can we say evolution is not about the conquest of new environments? It requires a good philospher to shed light on this question.

The pressure is more often on increased performance in a complex environment using highly complex foods and locomotions.

Increased complexity is an inevitable ramification of the evolutionary process through time and space, rather than a direct and inevitable requirement of it.

Because species evolve into new niches, The most logical and efficient way to do that, is to keep the genes for old niches, in the DNA library, and to add new ones next to it. If the organism did not keep genes from old niches, and use them for a proportion of it's mutations, it would be less apt. Useful genes are costly, they can cost millions of years to find them, for example The more of a toolbox of biochemistry and morphology.

For Biochemistry, life "discovers" new materials and proteins, and puts them to use, and it keeps a record of those materials after they are not needed.

A sea slug can evolve into a vertebrate fish, but a fish can't evolve back to a slug, because complexity enhances fitness, so perhaps we can say that fitness and complexity are not disassociable.

Change is complex thing, and evolution is about changing, so for me, evolution adds complexity every time it changes.

  • 2
    $\begingroup$ You need some citations here, especially because your answer is somewhat at odds with existing answers. It's also quite difficult to read. $\endgroup$ – Bryan Krause Jul 21 '17 at 20:40
  • 1
    $\begingroup$ Except life very often evolves to be simpler, vestigial and lost structures are a great example. complexity can enhance fitness it can also reduce fitness. $\endgroup$ – John Nov 1 '19 at 1:21

The simplest way to look at it is there is a near infinite number of ways to be more complex but a very limited number of ways to be simpler. There is even fewer ways of being simple. So even with just pure random variation, over time all things being equal you will end up with more complex organisms.

This becomes even more true when you take competitor into account, competition, go too simple and you lose the ability to do things that you really need to compete with the rest of life around you, go too simple and you can't reproduce fast enough to keep up with all the things eating you. While on the other hand the cost of complexity can be offset by better capabilities. On top of this imagine a ruler with as simple as life can be and still function on one side and as complex as life can get and still function on the other. the first life is going to be pretty close to the simplest possible life, so most of the ways of being alive that are possible are going to be more complex, so again even if you ignore selective pressures either way, just random variation is going to create more complex life than simple life. There is more complex phase space to occupy than simple phase space.

imagine I stand with my back to a cliff and throw a ball randomly in the air, now after I throw a thousand balls the vast number of balls I find are going to be in front of me, not becasue I am actively trying to throw them there but because most of the balls that fall behind be get lost, (go extinct for our analogy)

  • $\begingroup$ "The simplest way to look at it is there is a near infinite number of ways to be more complex but a very limited number of ways to be simpler. " I think this argument is flawed: it would mean that in artificial life simulations you would see the same tendency towards enhanced complexitiy... which is false. The agents in those simulations tend to stay on the same level of complexity. $\endgroup$ – vonjd Nov 1 '19 at 11:28
  • $\begingroup$ @vonjd I will admit simulating life is not my forte, but simulations of evolution do often show an increase in complexity, that is not to say only more complex life evolves but that on average there are more complex entities than your starting condition. Of course the actual history of life also shows this. I would argue not seeing increasing complexity in a simulation is evidence of the limitations of the simulation since it is not matching the real world system it is trying to simulate. $\endgroup$ – John Nov 1 '19 at 13:13

Not the answer you're looking for? Browse other questions tagged or ask your own question.