10 ms·

New Model of Evolution Reveals How Cooperation Evolves

83 points by RockyMcNuts 9y ago

sharemywin 9y ago
That would be a neat derivative of poker. Same as regular poker where everyone antes but if everyone folds they get 1.1 X their ante. if one person doesn't fold, you play it like regular poker.
- ftlio 9y ago
  Who pays out the .1?
  - andrewdb 9y ago
    Perhaps a rotating big and small blind on top of ante?
    - jessaustin 9y ago
      It seems Margaret Thatcher had something to say along these lines?
    - sharemywin 9y ago
      Maybe the person with the least amount of chips or the one with the most depending on whether you want the game to go longer or shorter.
trhway 9y ago
>a crucial factor turns out to be the process of punishment. “Punishment acts like a magnetic field that leads to an 'alignment' between players, thus encouraging cooperation,” say Adami and Hintze. either you're a team player or else. Collective punishment comes to mind too as a very effective way to induce cooperation/alignment inside the collective which is going to be or threatened to be punished.
- nilson 9y ago
  what about other cooperation methods beside punishment. why are they less crucial?
  - woodandsteel 9y ago
    Society already includes many forms of punishment, and there is no way that would ever change. The question is which forms of punishment are helpful and which not, and the social scientists can answer that question only if they include it in their studies.
- woodandsteel 9y ago
  Yes, they start out with game theory, which is only about rewards, and then add in punishment. Game theory is enormously insightful, but It leaves out a great many factors that are very important in the real world, such as communication between the parties,uncertain rules, and reputation.
  - jballanc 9y ago
    All the factors you mention are accounted for by researchers that use game theory as the basis for evolutionary modeling (for a primer, check out "Evolutionary games on graphs" - https://arxiv.org/abs/cond-mat/0607344 https://arxiv.org/abs/cond-mat/0607344).
    - woodandsteel 9y ago
      Ok, I was confusing the prisoner's dilemma with game theory as such.
ouid 9y ago
>But evolution is a random process based on the short-term advantages that emerge in each generation. This idea is easily discredited, even "disadvantageous-but-viable" mutations are prized possessions among my offspring. Environments change, and whether or not my lineage goes unbroken is a result of its ability to survive those changes. Those environmental changes are unpredictable. phenotypes that benefit me now do not necessarily benefit me in the future. If it is my goal to extend my lineage as long as possible, then my optimal strategy must give extra value to diversity. Since every organism possesses a lineage that goes back to the origin of life, it's fair to say that cooperation for the goal of increasing diversity is a pretty heavily selected for trait.
- onychomys 9y ago
  You're basically arguing for a species-level (or at least a deme-level) view of selection. That's something that we've tried to find for many decades, and there just isn't a lot of evidence for it. Selection happens at the level of the individual animal, not any higher grouping.
  - jawarner 9y ago
    Why doesn't Hamilton's rule or evolutionary game theory in general justify a social-network aware view of evolution? What evidence or theoretical argument would constitute a species level view of selection?
  - statevol 9y ago
    I've taught evolutionary biology to non-biology majors and the way I would explain natural selection is as follows. First, to clarify the last sentence above, selection does not happen at the level of an individual. Selection is a statistical process that requires a population of individuals. For evolution by selection to occur, there has to be 1) variation in traits (i.e. "Phenotypes"), 2) the variation in these traits has to be in part heritible (i.e. Have a genetic basis), and 3) there have to be average differences in success of leaving offspring that themseves go on to reproduce, etc, etc, based on the trait differences amongst individuas.
throwawaymanbot 9y ago
Remember this article for the next time somebody complains about how perfect an Ayn Randian world would be. Remember that free cooperation helped humanity get this far, not Rugged, Individualism.
- ende 9y ago
  First you'll have to teach those strawmen to read.
woodandsteel 9y ago
I wonder if this has any relevance to the question of the evolution of altruism.
- eevilspock 9y ago
  White blood cells are clearly altruistic. To the extent a species or even an ecosystem functions like an organism, there is an evolutionary advantage to evolving altruism. To quote Spock, "where the good of the many outweigh the good of the few, or the one."
- int_19h 9y ago
  There have been models that pretty convincingly show altruism arising without any special prerequisites. It's basically a balance between propagating directly versus ensuring propagation of enough of your kin. They may even be very distant kin, not sharing that many genes with you - but if your altruism helps sufficiently many by sufficiently much, you end up propagating those shared genes more successfully, in a sense of having more carriers of them in the follow-up generations. And if those shared genes include the ones that affected your altruistic behavior - again, in at least some of those you helped, not necessarily all - then it becomes a selectable trait.
dhfhduk 9y ago
Cool take on things, but not really fundamentally new as an explanation of evolution of cooperative traits. Models involving reputation and punishment have been around for awhile. Basically, once you introduce reputation and memory, together with punishment, cooperation becomes more viable as a strategy. It's even led to models of higher-order cooperation, of enforcing the enforcers, etc. To me it seems like the paper isn't really offering a fundamentally new explanation of how these traits arise, it's just providing a sophisticated analysis of the dynamics.
- basch 9y ago
  I remember reading this article 7 years ago http://seedmagazine.com/content/article/the_evolution_of_cooperation/ http://seedmagazine.com/content/article/the_evolution_of_coo...
- aqsalose 9y ago
  >To me it seems like the paper isn't really offering a fundamentally new explanation of how these traits arise, it's just providing a sophisticated analysis of the dynamics. MIT Tech Review has often news about fascinating research, but one has to adjust for some amount of unnecessary hype.
- jballanc 9y ago
  That cooperation is eventually selected for has been known for decades, but we didn't know how. Yes, there have been many speculative models describing how cooperation might arise, kin selection being among the earliest of these. Each of these mechanisms has some applicability and some predictive power (for example, kin selection is pretty good at describing cooperation in everything up to and including eusocial insects), but we've been missing the general mechanism that has predictive power across all domains. In other words, the field of evolutionary biology today has its Newton's laws and Maxwell equations, but we're missing anything like the Standard Model. The authors of this work have been at this for a number of years. A sophisticated analysis of the dynamics is a key and very important step in beginning to develop a "grand unified theory" of evolution.
lngnmn 9y ago
Cooperation is related to a family (or a clan/pack) life. It is the "most natural" survival strategy. Expanding it to a non-keen is not a big deal. Nothing to see here.
jballanc 9y ago
Almost all of the responses here putting this down as "not that big a deal" are completely missing the point. It's extremely easy to say that evolution should do a thing, but if you cannot provide a mechanism to explain how such a thing is passively selected for, then evolution doesn't care. Evolution has no intelligence. There is no designer. This is probably the most important aspect of evolution to understand and the one hardest to grasp intuitively. For example: say you had a savannah with grass and tall trees populated by 4-legged horse-like animals that survive by grazing on the grass. We can look at such a scenario and say that it would be advantageous for some sub-population of those horse-like animals to grow longer and longer necks until they were capable of also grazing on the trees (doubling their available food supply, relative to their neighbors), but it will never happen! In each generation there will be a distribution of neck lengths, and as there is no advantage to those with slightly longer (but not long enough to reach the trees) necks, the distribution will remain unchanged. This is the basic Hardy-Weinberg principle. The only way you get giraffes is if the trees start out much shorter, so that the horse-like animals are capable of grazing on them. Then the trees that are slightly taller will have an advantage, and the distribution of tree heights in the population will shift. This will give an advantage to the horse-like animals with longer necks, causing their distribution to shift, and on, and on. This is the basic Red Queen hypothesis. So saying that an organism that has an inherent capacity for adaptation, or one that knows to cooperate with neighbors, would be better able to survive than the alternatives doesn't mean crap. If you cannot provide a mechanism (backed by a mathematical model) of how such traits arise, then evolution doesn't care. That is what makes this work so particularly interesting. The authors are providing a model to explain how a mechanism for the selection of cooperation could occur.
glup 9y ago
Dear science journalists, please quit saying that papers "solve" problems ("By treating evolution as a thermodynamic process, theorists have solved one the great problems in biology."). Also, as much as I love Arxiv, I wish journalists would note that articles posted are not subject to peer review.

83 points by RockyMcNuts 9y ago

That would be a neat derivative of poker. Same as regular poker where everyone antes but if everyone folds they get 1.1 X their ante. if one person doesn't fold, you play it like regular poker.

ftlio 9y ago
Who pays out the .1?
- andrewdb 9y ago
  Perhaps a rotating big and small blind on top of ante?
  - jessaustin 9y ago
    It seems Margaret Thatcher had something to say along these lines?
  - sharemywin 9y ago
    Maybe the person with the least amount of chips or the one with the most depending on whether you want the game to go longer or shorter.

trhway 9y ago

>a crucial factor turns out to be the process of punishment. “Punishment acts like a magnetic field that leads to an 'alignment' between players, thus encouraging cooperation,” say Adami and Hintze. either you're a team player or else. Collective punishment comes to mind too as a very effective way to induce cooperation/alignment inside the collective which is going to be or threatened to be punished.

nilson 9y ago
what about other cooperation methods beside punishment. why are they less crucial?
- woodandsteel 9y ago
  Society already includes many forms of punishment, and there is no way that would ever change. The question is which forms of punishment are helpful and which not, and the social scientists can answer that question only if they include it in their studies.
woodandsteel 9y ago
Yes, they start out with game theory, which is only about rewards, and then add in punishment. Game theory is enormously insightful, but It leaves out a great many factors that are very important in the real world, such as communication between the parties,uncertain rules, and reputation.
- jballanc 9y ago
  All the factors you mention are accounted for by researchers that use game theory as the basis for evolutionary modeling (for a primer, check out "Evolutionary games on graphs" - https://arxiv.org/abs/cond-mat/0607344 https://arxiv.org/abs/cond-mat/0607344).
  - woodandsteel 9y ago
    Ok, I was confusing the prisoner's dilemma with game theory as such.

ouid 9y ago

>But evolution is a random process based on the short-term advantages that emerge in each generation. This idea is easily discredited, even "disadvantageous-but-viable" mutations are prized possessions among my offspring. Environments change, and whether or not my lineage goes unbroken is a result of its ability to survive those changes. Those environmental changes are unpredictable. phenotypes that benefit me now do not necessarily benefit me in the future. If it is my goal to extend my lineage as long as possible, then my optimal strategy must give extra value to diversity. Since every organism possesses a lineage that goes back to the origin of life, it's fair to say that cooperation for the goal of increasing diversity is a pretty heavily selected for trait.

onychomys 9y ago
You're basically arguing for a species-level (or at least a deme-level) view of selection. That's something that we've tried to find for many decades, and there just isn't a lot of evidence for it. Selection happens at the level of the individual animal, not any higher grouping.
- jawarner 9y ago
  Why doesn't Hamilton's rule or evolutionary game theory in general justify a social-network aware view of evolution? What evidence or theoretical argument would constitute a species level view of selection?
- statevol 9y ago
  I've taught evolutionary biology to non-biology majors and the way I would explain natural selection is as follows. First, to clarify the last sentence above, selection does not happen at the level of an individual. Selection is a statistical process that requires a population of individuals. For evolution by selection to occur, there has to be 1) variation in traits (i.e. "Phenotypes"), 2) the variation in these traits has to be in part heritible (i.e. Have a genetic basis), and 3) there have to be average differences in success of leaving offspring that themseves go on to reproduce, etc, etc, based on the trait differences amongst individuas.

throwawaymanbot 9y ago

Remember this article for the next time somebody complains about how perfect an Ayn Randian world would be. Remember that free cooperation helped humanity get this far, not Rugged, Individualism.

ende 9y ago
First you'll have to teach those strawmen to read.

woodandsteel 9y ago

I wonder if this has any relevance to the question of the evolution of altruism.

eevilspock 9y ago
White blood cells are clearly altruistic. To the extent a species or even an ecosystem functions like an organism, there is an evolutionary advantage to evolving altruism. To quote Spock, "where the good of the many outweigh the good of the few, or the one."
int_19h 9y ago
There have been models that pretty convincingly show altruism arising without any special prerequisites. It's basically a balance between propagating directly versus ensuring propagation of enough of your kin. They may even be very distant kin, not sharing that many genes with you - but if your altruism helps sufficiently many by sufficiently much, you end up propagating those shared genes more successfully, in a sense of having more carriers of them in the follow-up generations. And if those shared genes include the ones that affected your altruistic behavior - again, in at least some of those you helped, not necessarily all - then it becomes a selectable trait.

dhfhduk 9y ago

Cool take on things, but not really fundamentally new as an explanation of evolution of cooperative traits. Models involving reputation and punishment have been around for awhile. Basically, once you introduce reputation and memory, together with punishment, cooperation becomes more viable as a strategy. It's even led to models of higher-order cooperation, of enforcing the enforcers, etc. To me it seems like the paper isn't really offering a fundamentally new explanation of how these traits arise, it's just providing a sophisticated analysis of the dynamics.

basch 9y ago
I remember reading this article 7 years ago http://seedmagazine.com/content/article/the_evolution_of_cooperation/ http://seedmagazine.com/content/article/the_evolution_of_coo...
aqsalose 9y ago
>To me it seems like the paper isn't really offering a fundamentally new explanation of how these traits arise, it's just providing a sophisticated analysis of the dynamics. MIT Tech Review has often news about fascinating research, but one has to adjust for some amount of unnecessary hype.
jballanc 9y ago
That cooperation is eventually selected for has been known for decades, but we didn't know how. Yes, there have been many speculative models describing how cooperation might arise, kin selection being among the earliest of these. Each of these mechanisms has some applicability and some predictive power (for example, kin selection is pretty good at describing cooperation in everything up to and including eusocial insects), but we've been missing the general mechanism that has predictive power across all domains. In other words, the field of evolutionary biology today has its Newton's laws and Maxwell equations, but we're missing anything like the Standard Model. The authors of this work have been at this for a number of years. A sophisticated analysis of the dynamics is a key and very important step in beginning to develop a "grand unified theory" of evolution.

lngnmn 9y ago

Cooperation is related to a family (or a clan/pack) life. It is the "most natural" survival strategy. Expanding it to a non-keen is not a big deal. Nothing to see here.

jballanc 9y ago

Almost all of the responses here putting this down as "not that big a deal" are completely missing the point. It's extremely easy to say that evolution should do a thing, but if you cannot provide a mechanism to explain how such a thing is passively selected for, then evolution doesn't care. Evolution has no intelligence. There is no designer. This is probably the most important aspect of evolution to understand and the one hardest to grasp intuitively. For example: say you had a savannah with grass and tall trees populated by 4-legged horse-like animals that survive by grazing on the grass. We can look at such a scenario and say that it would be advantageous for some sub-population of those horse-like animals to grow longer and longer necks until they were capable of also grazing on the trees (doubling their available food supply, relative to their neighbors), but it will never happen! In each generation there will be a distribution of neck lengths, and as there is no advantage to those with slightly longer (but not long enough to reach the trees) necks, the distribution will remain unchanged. This is the basic Hardy-Weinberg principle. The only way you get giraffes is if the trees start out much shorter, so that the horse-like animals are capable of grazing on them. Then the trees that are slightly taller will have an advantage, and the distribution of tree heights in the population will shift. This will give an advantage to the horse-like animals with longer necks, causing their distribution to shift, and on, and on. This is the basic Red Queen hypothesis. So saying that an organism that has an inherent capacity for adaptation, or one that knows to cooperate with neighbors, would be better able to survive than the alternatives doesn't mean crap. If you cannot provide a mechanism (backed by a mathematical model) of how such traits arise, then evolution doesn't care. That is what makes this work so particularly interesting. The authors are providing a model to explain how a mechanism for the selection of cooperation could occur.

glup 9y ago

Dear science journalists, please quit saying that papers "solve" problems ("By treating evolution as a thermodynamic process, theorists have solved one the great problems in biology."). Also, as much as I love Arxiv, I wish journalists would note that articles posted are not subject to peer review.