Thursday 22 June 2017

Nowak and Wilson are at it again in 2017

Nowak and Wilson are at it again, with a 2017 paper:

"The general form of Hamilton’s rule makes no predictions and cannot be tested empirically"

It's the old claim that "survival of the fittest" is a tautology (since the fittest are those that survive) repackaged for kin selection.

Evolutionary biologists have long had a stock answer to the claim that "survival of the fittest" is a tautology. This hinges on the difference between expected fitness and observed fitness.

"Survival of those with the highest observed fittesses" would indeed be a kind of circular argument. However, "survival of those with the highest expected fittesses" is making a falsifiable statement about the model with which the expected fitnesses are calculated.

Nowak and Wilson say:

The mathematical investigation of HRG reveals three astonishing facts. First, HRG is logically incapable of making any prediction about any situation because the benefit, BB, and the cost, CC, cannot be known in advance. They depend on the data that are to be predicted. At the outset of an experiment, BB and CC are unknown, and so there is no way to say what Hamilton’s rule would predict.

Those are observed fitnesses (measured after the fact). Rather obviously, if you want to make predictions, you need to be using expected fitnesses when determining B and C. Nowak and Wilson are simply mistaken in claiming that there's no way in which these could possibly be calculated in advance. You could predict these values by using a mathematical model of the situation, or by using a computer simulation - for example.

IMHO, it is pretty amazing that some scientists bothered to write a whole paper about this issue, without first understanding it.

Wednesday 28 September 2016

Group selection vs kin selection: evidence from examples

Modern versions of kin selection and group selection have turned out to be essentially equivalent. However, historical examples of his selection and group selection have covered some rather different territory. Group selection enthusiasts have chosen phenomena such as senescence and warning signals. From the perspective of kin selection, some of the most obvious examples include breast feeding, parental care and nepotism. The kin selection examples are good, but the group selection ones are more dubious. I've previously argued that the reason group selection was associated with dodgy examples is because advocates sought to distinguish their theory from kin selection - and so avoided examples where it was obvious that kin selection was responsible. The details are in my 2104 article "How kin selection pushed group selection into the scientific fringes".

I think we are seeing a similar phenomenon with cultural group selection and cultural kin selection. My articles on cultural kin selection are illustrated with examples of where kin selection works on shared memes, rather than shared genes. These examples include uniforms, money and religious brotherhoods and sisterhoods. I think these are real examples of cases where Hamilton's rule applies to memes. However, the cultural group selection examples look rather different. One example often given involves Dinka and Nuer populations in the Sudan. Another example involves monogamous marriage customs. In these cases, a simple explanation is that some memes involved are fitter than other ones. Evidence that the memes involved are deleterious to individuals yet spread through group level reproduction or extinction seems to be completely missing. Maybe these are just bad memes. Cultural group selection seems largely unnecessary.

The tendency of group selection enthusiasts to pick bad examples seems common to both the organic realm and the cultural realm. However the explanation I have previously given for the bad examples in the organic realm does not seem to be applicable to the cultural realm. In the case of cultural evolution, cultural group selection seems to be much better known than cultural kin selection is. So, the idea that cultural group selection was bent out of shape to avoid cultural kin selection does not seem very plausible. Another explanation would seem to be required.

My perception is that the cultural group selection enthusiasts are choosing examples of cases where one group does better than another one for cultural reasons. However, that's not a particularly good way of finding examples of where Hamilton's rule applies. The native Americans did badly as a group recently - but that wasn't really due to kin selection or group selection. Their environment changed rapidly and they failed to adapt quickly. Hamilton's rule isn't terribly relevant in cases like these - even though one group is exterminating another one. This is part of the historical problem with group selection. People see groups apparently exterminating each other and then leap to group selectionist explanations without bothering to properly test them to see if they are plausible.

One possible reply by group selection advocates is that kin selection and group selection don't cover the same ground after all. However, lots of people have held that position over the years, but the evidence for it doesn't seem to be there. If you accept equivalence, I think is is hard to deny the prevalence of bad quality examples of group selection. When was the last time you heard breastfeeding or parental care given as an example of group selection? Never, right? Group selection advocates are not thinking about mother-offspring pairs as "groups" - even though that's where the maths says the effect is strongest. Inferior examples associated with group selection arise in both the cultural and organic realms. My previous argument about how kin selection pushed group selection into the scientific fringes doesn't look as though it is the whole explanation. Maybe group selection is just misleading or confusing.

Saturday 10 September 2016

The Meaning of Human Existence (not)

I read chapter 6 of Edward O Wilson's "The Meaning of Human Existence" from 2014.

This is the chapter on 'inclusive fitness theory'. It was a jaw-dropping experience. If we are to believe Edward's account, he helped to lead a valiant crusade against a long-established theory and succeeded in overturning it in 2010. Of course, the reality is that the 2010 paper in question was an embarrassing scientific farce, which had no useful new results and just paraded the authors' own misconceptions.

I think that the lesson here is that the egos of famous elderly scientists can get out of control, leading to delusions of grandeur. Other famous elderly scientists should take note: don't let this happen to you.

Tuesday 16 February 2016

Google NGRAM for kin selection

Here's the Google NGRAM for group selection, kin selection, inclusive fitness, multilevel selection. It is interesting, I think.

Monday 28 December 2015

Answers for Doug Hoxworth

Doug Hoxworth had some questions which he aired in public recently:

  • If transmission only happens based on the “selfish” gene theory, how do behavioral or personality traits that are advantageous for the group but disadvantageous for the individuals within the group ever get transmitted to the next generation and become dominant/ubiquitous? If it is heritable, how can this phenomenon be explained?

    • Kin selection. For example, worker sterility is disadvantageous to the worker as an individual, but beneficial to the group they are part of. The reason for this is because the worker and the reproductives (queens and drones) share genes. It is possible to explain this in terms of group selection as well. However group selection explanations often turn into simplistic group-level functionalism - whereas kin selection quantifies relatedness (using 'r') and so is a better tool for making quantitative predictions with - in those cases where relatedness is less than 1.

  • How can altruism to non-relatives be explained?

    • For humans, the most significant mechanisms are probably cultural kin selection, virtue signaling, manipulation, over-generalization and environmental mismatch - in roughly that order.

  • If it is through enculturation and imitation, how are these behaviors so ubiquitous at a very young age, if not at birth (even applicable to animals/organisms other than humans that presumably do not emit pheromones indicating that they are unambiguously related, e.g., ants)?

    • Cultural transmission is often advantageous to individuals. Using socially-transmitted information from others can give a short-cut to learning - if maladaptive traditions can be avoided. There are various ways of avoiding maladaptive traditions.

Doug's article contrasts group selection with the selfish gene. In fact these ideas are compatible. The selfish gene is compatible with kin selection and it is compatible with group selection. Richard Dawkins might not agree - but we don't need to heed him here. The disagreement in this area mostly lies elsewhere.

The problem with group selection is not so much that it's wrong, but that it causes confusion among its practitioners. For decades group selection advocates held out hope that their theory would make novel predictions. In the last decade, this hope has mostly collapsed and most now recognize that group selection and kin selection are broadly equivalent. It is a matter of different accounting techniques, so to speak.

The main problem is that we know that kin selection is strongest between close relatives. Group selection advocates often want to apply the theories to whole tribes and to warfare. This sort of group selection isn't equivalent to kin selection - and, for the most part, it doesn't actually work.

What is left of this whole debate? Not too much. There's some noise surrounding Hamilton's inclusive fitness concept, but this is mostly coming from Martin Nowak and friends - and his papers have been met with ridicule. At this stage, most of the facts seem to be in and the group selection controversy resembles a mopping up operation.

Sunday 23 August 2015

Prospective kin

Organisms treat their mates better than average conspecifics - even before they have had any offspring with them.

It isn't hard to think of reasons for this, but many of of those reasons invoke the idea of prospective kin - unborn children.

Most children have a probability of becoming an ancestor - and it is possible to extend this idea to children that haven't been born - and to children that haven't been even conceived.

Pretty standard kin selection models can thus be applied to courtship behaviour, bower construction, nuptual gifts - and so forth. These are cases where cooperative behaviour without relatedness occurs. In fact, there is relatedness - relatedness to unborn children.