Genes leading to wars (simple model)

From: Keith Henson (
Date: Thu 02 Sep 2004 - 13:37:29 GMT

  • Next message: Vincent Campbell: "contagious terrorist methods"

    As I have discussed here before, this is a model where anticipation of hard times (originally *starvation*) turned up the gain in human groups of xenophobic memes. After enough time for the memes to dehumanize the tribe in the next valley, and work the warriors up to a killing frenzy, the human population (in stone age times) was reduced. As long as the future looked bleak they went at each other. When prospects started to look better, the tribes quit fighting and put more effort into raising children for the
    *next* cycle.

    It's a case where humans have to become their own predators in times they are needed because a long time ago our ancestors became *top* predator and the big cats no longer were up to controlling hominid numbers. (Something must limit populations.)

    I was recently asked to provide some rough modeling on another list. I did but there were no replies (so far). Since I claim that memes are in the causal loop here, this is appropriate subject matter for this list as well.


    >Keith Henson wrote:
    >>PS. Wars are *adaptive* for hunter gatherer societies. That does not
    >>make necessarily make wars adaptive to later societies.
    >I am not sure this follows from the fact that hunter-gatherer societies
    >war. It could be the equivalent of a mutual defection in the Prisoner's
    >Dilemma; not adaptive when compared to mutual cooperation, but still a
    >Nash equilibrium.

    That's certainly a point to consider. But in an environment with no birth control, and a fixed technology base where chipped rocks are high tech, I don't see what form mutual cooperation might take. Any thoughts?

    >I do not find convincing, as words on paper, the argument that the losing
    >tribe's women and children would be absorbed and that this is preferable
    >to mutual starvation.

    Darn near anything that does not kill everybody concerned is better than universal starvation.

    >I want to see a mathematical model describing the necessary conditions for
    >this, or a study of reproductive outcomes in hunter-gatherer tribes.

    Serious fluctuations in the carrying capacity of the ecosystem happening once ever few years to once every few generations would be enough and certainly within the range we have seen in the tree ring and other records.

    >There is no obvious genetic reason for a tribe to adopt male children,

    Generally they were not, though there were exceptions where an awful lot of the tribe's males had been killed off.

    >and as I recall my childhood Torah studies, "God" told the Israelites to
    >kill off all conquered tribesfolk except virgin girls. That's not many
    >surviving genes.

    Better than none.

    Taking a typical episode, consider a fluctuation where two equal sized tribes occupy an area where there is a food reduction where only 75% of them would live through it.

    Assume that tribal behavior is the outcome of genes A and S for attack vs starve, and the tribe takes the course of the majority. Further, let's assume that before this episode, the genes are close to a 50/50 distribution. (S type tribe 45/55 split, A type 55/45 split, first attacker type 60/40 A/S ratio.) It is also reasonable to assume that being attacked is enough to get even a tribe with a slight majority of S genes to retaliate.

    There are three conditions, both S, both A, and one S the other A. There is the well known factor that surprise attacks are more effective, the tribe with the higher A percentage attacks first and gets some advantage, but the outcome is not highly biased, say 60/40 for win/lose if you attack first. Assume the young women carry the A and S genes in the same ratio as the males and that the next generation of the mixed tribe carries S and A genes in the same proportions.

    Both S, both tribes take a 25% hit in numbers, equal male and female deaths. The tribes will be slower to rebound when the ecosystem becomes more productive.

    Both A, the tribe with the higher numbers of A attacks first and has a 60 percent chance of winning. For simplicity the winning tribe kills the losing males without loss. Since the food and population ratio has been restored, all the females and enough males get through the hard times. The numbers work that the next generation has about the same A to S ratio.

    .6 x (.60 (1) + .55(.5) /1.5 = .35


    .4 x (.60 (.5) + .55(1)/1.5 = .27

    So given these numbers the A genes have crept up to .62, not much but in the right direction.

      One S and one A.

    in the case the S tribe wins (40%),

    .45(1) + .55(.5)/1.5 = .4833 (.1933)

    if they lose (60%)

    .45(.5) + .55(1)/1.5 = .5166 (.31)

    So on average the resultant would still be a tribe slightly over attack threshold.

    So as long as there are collections of tribes with A genes slightly in the lead, the next generation will be more prone to war.

    The factor that might make this go quicker is the size of the next generation. Tribe pairs that had collectively killed half the males to get through the hard times would bounce back faster with twice as many women.

    Now that I have been thinking about this, it is known that other hominids, up to and including Neanderthals died out in contact with our kind of human. Wars of this kind are, I make the case, based on anticipation of bleak prospects. It could be that the ability to anticipate future hardships (because the game had mostly been killed or the berry crop failed) was unique to our line and induced us to kill not only fellow members of our species when hard times were a-coming, but other species in the same ecological niche.

    Keith Henson

    PS. The check to prevent this from running away is that being too prone to fighting when not justified by a worse alternative is not conducive to getting your genes into the next generation.

    =============================================================== This was distributed via the memetics list associated with the Journal of Memetics - Evolutionary Models of Information Transmission For information about the journal and the list (e.g. unsubscribing) see:

    This archive was generated by hypermail 2.1.5 : Thu 02 Sep 2004 - 13:56:23 GMT