A Query for my Scientifically Oriented Friends

Comments

4 responses to “A Query for my Scientifically Oriented Friends”

1. 

   July 15, 2009

   Lifewish

   Yup, it’s daft. “Social” behaviours can emerge in any situation where

   1) members of species A are distinguishable
   2) members of species A rely on species B for resources
   3) members of species B may withhold those resources
   4) members of species B have a good memory

   Species A and species B may be the same species, but it’s not essential.

   For example, consider grouper fish and cleaner fish. The groupers let the cleaners remove bits of carcass from between their teeth. This is mutually beneficial – the cleaner gets food, the grouper gets dental hygiene.

   But what’s to stop the grouper just waiting until the cleaner has done its thing and then chowing down on it before it can escape? This would be bad for groupers in general (fewer cleaners = bad breath) but good for the individual. And in general evolution works based on what’s good for the individual not the species.

   It turns out that this does occasionally happen – groupers “go rogue” and start snacking on cleaners. But the other cleaners take note of this and avoid that particular grouper, which then has to deal with its girlfriend making snide remarks about breath mints.

   Other behaviours are possible that, to the casual eye, look like altruism. For example, a worker ant will selflessly resist the urge to lay eggs, instead focusing on looking after her queen’s offspring.

   Or will she?

   It turns out that it’s entirely possible for workers to lay fertile eggs. The catch? If they try to pull a stunt like that, the eggs get attacked and eaten by other workers.

   The problem is that, in general, a female ant will be more closely related to her mother than to her sisters. So her genes are more likely to live on in her mother’s eggs than her sisters’ eggs.

   Of course the best thing would be for her to be laying the eggs herself. But if it’s a choice between her mother having kids and her sisters having kids, the former is definitely the better option. Better to kill off her sisters’ offspring, then, so they’ll stay focused on looking after her mother’s babies.

   In effect, then, all the workers in any given hive are perpetually ganging up on each other. The result is that only the queen can have babies. This gives the appearance of altruism, despite actually being infanticide (insecticide?).

   Being a cynical sort, I suspect that this is closer to what’s happening in queues. If you push in, you gain a small amount of advantage, but you also vastly increase the chance that everyone else in the queue will get angry and beat you to a pulp. Usually it’s not worth the risk.

   These anecdotes are straight off the top of my head. I think I nicked them both from books by Matt Ridley, but it’s been a while. So take with a pinch of salt.

   Reply
   1. 

      July 15, 2009

      Lifewish

      Right, having now browsed the press release and the primary author’s webpage, I think I can decipher what’s going on.

      I can’t read the paper itself because it’s not printed til September, so the following is an educated guess based on the author’s other papers and my own small knowledge of evolutionary game theory. Apologies for the length,

      It looks like the authors have done a perfectly acceptable bit of research. What they’re looking at is similar to what I did with my four-point list in the above post: they’re trying to establish the minimal conditions under which certain behaviours can arise.

      The specific behaviour they’re analysing here is the ability for two critters to take turns to their mutual benefit, and without exposing themselves to freeloaders. They’re analysing this problem using “evolutionary game theory”, which is a field that uses simple abstract games to study how strategies can evolve over time.

      An example. Let’s say that your web development company provides free donuts. It leaves a big box of them in the canteen, and makes sure it never gets empty.

      Now, when you feel peckish, you can go grab yourself a donut. You can also grab donuts for everyone else. This takes a bit of extra effort, but if everyone does that then no-one has to make as many canteen trips.

      There’s a problem. Your colleague Steve is not a very pleasant person, and he never bothers to go get donuts. He prefers to just wait until someone else gets some, and then scrounge off them. How do you adjust (“evolve”) your strategy to deal with this?

      The simplest way is by saying: I won’t get people donuts unless they’ve got me one recently. Sorry, Steve, but until you pull your weight we’re not feeding you any more. This is called the “tit for tat” strategy, and in most cases involving reciprocal altruism it is the least bad approach: no-one can freeload off you for long.

      Unfortunately, your company has been hit hard by the recession. They don’t have as much money for donuts any more. So they decide they’ll only replace the donuts every hour. This being an IT company, the effect is akin to dropping a cow in a piranha pool.

      At 9:00 Monday morning, you’re feeling peckish so you decide to go up to grab some donuts. But everyone else in your team is also peckish, so they do the same. And since you’re all IT professionals, no-one bothers to communicate their plans to anyone else. The result is that, by 9:05, everyone has a desk piled high with donuts and no-one to share them with.

      Now, normally this would be a good problem to have… But then 10:00 comes along. And you think to yourself “well, I could get donuts for the entire group, but I don’t remember any of the lazy bastards giving me one at 9:00. Better just get one for myself then.” This is a bit daft of you – it wasn’t like you needed the extra donuts earlier – but you’re determined to stick with your tit-for-tat rule.

      Of course, everyone else is thinking the same thing, so everyone just gets donuts for themself. The same thing happens at 11:00, and 12:00, and so on. Co-operation breaks down completely and suddenly everyone is spending five minutes an hour walking to and from the donut box. Productivity plummets. Darn those donuts!

      Fortunately, the following week, a new graduate arrives in your group. The grad has, of course, spent their entire university career getting out of bed at lunchtime. As a result, their metabolism is artificially slow in the morning – they don’t get peckish until much later than everyone else.

      So 9:00 rolls round and the same thing happens as last week: everyone goes to get donuts (except the grad). And rather than having to eat all your donuts, you can now donate one of your spares to the newbie (whose appetite has picked up at the sight of all that sugar). When 10:00 rolls round, the grad reciprocates by getting everyone a donut. At 11:00, everyone else re-reciprocates, and so on. Co-operation is restored.

      The implication is that, for tit-for-tat strategies to work in turn-based situations, you need a certain amount of initial diversity in approach. This sounds incredibly obvious, and to an extent it is, but there’s a lot of value in quietly proving this sort of very basic principle.

      The only problem is when the university press office gets hold of your research and decides to make it more “relevant”. They start throwing around lots of cute analogies – like the queuing one – that aren’t remotely relevant to your work. They take your quotes out of context, so people think you’re saying that the evolution of altruism is somehow rare or implausible.

      And once the mainstream media get their teeth into it, it’s all over bar the swearing. The moral: information is only as good as the route it took to get to you.

      Reply
      1. 

         July 15, 2009

         Lifewish

         Curse you, missing preview button!

         Reply
         1. 

            July 15, 2009

            Henry Neufeld

            I fixed it. And yes, I need to fix the theme as well, but it’s on a list of about a thousand web things I need to do.

            Reply