Sunday, 30 September 2012

Kin selection without brains

One of the common misunderstandings of kin selection is that recognising relatives requires advanced cognition:

In passing it needs to be remarked that the epistemological problems presented by a lack of linguistic support for calculating, r, coefficients of relationship, amount to a serious defect in the theory of kin selection. Fractions are of very rare occurrence in the world’s languages, appearing in Indo-European and in the archaic civilizations of the Near and Far East, but they are generally lacking among the so-called primitive peoples. Hunters and gatherers generally do not have counting systems beyond one, two and three. I refrain from comment on the even greater problem of how animals are supposed to figure out how that r [ego, first cousins] = 1/8. The failure of sociobiologists to address this problem introduces a considerable mysticism in their theory.

- Sahlins (1977) The Use and Abuse of Biology: An Anthropological Critique of Sociobiology.

Please give me an example where kin selection has been conclusively demonstrated in any species other than animals. I'm not interested in theoretical speculations or models. I'm interested in actual evidence.

- Moran, Larry (2012) Squirrels, Dawkins, and Evolution

Er, kin selection is not confined to animals. See, for example:

Equivalence naysayers

Not everyone is on board with kin selection and group selection turning out to be the same thing. Some quotes from the naysayers:

The non-equivalence of group selection and kin selection is therefore not only an important finding in itself, but also a case where the use of the Price equation leads to a claim that is not correct.

- Van Veelen M, García J, Sabelis MW, Egas M. (2012) Group selection and inclusive fitness are not equivalent; the Price equation vs. models and statistics.

Moreover, the claim that group selection is kin selection is certainly wrong.

- Nowak, Martin A., Tarnita, Corina E., and Wilson, Edward O. (2010) THE EVOLUTION OF EUSOCIALITY.

The concept of Group Selection has no useful role to play in psychology or social science.


Group selection', even in the rare cases where it is not actually wrong, is a cumbersome, time-wasting, distracting impediment to what would otherwise be a clear and straightforward understanding of what is going on in natural selection.

- Dawkins, Richard (2012) "Group Selection" Is A Cumbersome, Time-Wasting Distraction

He treated kin selection as a special case of group selection, an error which I was later to highlight in my paper on “Twelve Misunderstandings of Kin Selection” as Misunderstanding Number Two. Kin may or may not cling together in a group. Kin selection works whether they do or not.

- Dawkins, Richard (2012) The descent of Edward Wilson

the kin selection model is merely a special case of the multilevel selection theory
- Turchin, Peter (2011) Warfare and the Evolution of Social Complexity: A Multilevel-Selection Approach

Maybe, neither inclusive fitness nor evolutionary game theory can cover all mathematical intricacies of the other approach.

Ever since the publication of G. C. Williams' 1966 classic Adaptation and Natural Selection, biologists have joined with social scientists to form an altruism debunkery society. Any human or animal act that appears altruistic has been explained away as selfishness in disguise, linked ultimately to kin selection (genes help copies of themselves), or reciprocal altruism (agents help only to the extent that they can expect a positive return, including to their reputations). But in the last few years there's been a growing acceptance of the fact that "Life is a self-replicating hierarchy of levels," and natural selection operates on multiple levels simultaneously, as Bert Hölldobler and E. O. Wilson put it in their recent book, The Superorganism.
- Jonathan Haidt (2012) Contingent Superorganism

bees, ants and termites did not evolve their social behavior by group selection, but by a different mechanism known as kin selection
- Massimo Pigliucci (2011) Jonathan Haidt does it again, unfortunately

The altruism that evolves by group selection is "genuine" because it entails real self- sacrifice, while the altruism that evolves by kin selection is only "apparent" because it is just genes promoting copies of themselves in other individuals.


Multilevel selection is gaining in favor among evolutionary biologists because of a recent mathematical proof that kin selection can arise only under special conditions that demonstrably do not exist, and the better fit of multilevel selection to all of the two dozen known animal cases of eusocial evolution.
- E. O. Wilson (2013) The Riddle of the Human Species

To his credit, David Sloane Wilson has subsequently reversed his position on this issue.

However, as of 2015, he still denies that kin selection and group selection cover the same turf. In Challenge To Kin Selectionists. Explain This! he claims that group selection is a more general theory.

Modern kin selection and group selection concepts have turned out to have an enormous overlap. Even if they are not identical, they are certainly very close concepts.

Saturday, 29 September 2012

Similarity selection

Both Price's formalism and tag-based cooperation strongly suggest that it is trait similarity that is important to evolving cooperation - and that relatedness isn't the sole cause of similar identifying markers leading to cooperation.

This suggests that kin selection's name is dubious - and that the actual effect might be better described as a form of "similarity selection".

Reviewing the causes of similarity in biology, we find:

  • Similarity based on selection
    • Similarity based on mimicry
    • Similarity based on convergent evolution
  • Similarity based on chance
  • Similarity based on inheritance
    • Similarity based on organic inheritance
    • Similarity based on cultural inheritance
    • Similarity based on environmental inheritance
However, looking at where cooperation arises in nature, we see it between relatives, but there's relatively little sign of cooperation based on mimicry, convergent evolution or chance resemblances.

There are some examples of mimicry leading to cooperation. For example, cuckoos generate cooperative behaviour in their hosts by employing egg mimicry. Some orchids mimic the abdomens of their pollinators, to encourage insects to attempt to have sex with their flowers. These examples are instructive: mimicry seems to be associated with producing cooperative behaviour via deception and manipulation.

Chance and convergent evolution seem to result in cooperation much less frequently. The resulting resemblances are rarely close enough.

Convergent evolution may explain the similarity between the wings of bats and birds - however, neither the bats not the birds regard each other as kin, and they don't cooperate with each other especially frequently. Much the same goes for sharks and whales - which are another case of convergent evolution - but not much cooperation arises from the resulting resemblance.

The best example I can think of for similarity which is not caused by some kind of relatedness resulting in cooperation arises in a variant of the prisoner's dilemma. Imagine a group of unrelated agents of various types playing an iterated prisoner's dilemma game with each other - under circumstances where both agents can choose whether to continue with their current partner or pick another one at random from the pool of agents without partners. No agents reproduce, but the worst agents may die. In such a game, the "nice" agents will gradually find "nice" partners, and stick with them - gaining higher payoffs than other agents. They will do this regardless of whether they are related by kinship, or not. These kind of dynamics might help to explain some forms of mutualism. However, there's a problem with describing the cooperation in such models in terms of kin selection. An alternative explanation for the resulting cooperation is reciprocity.

Whether similarity selection really adds anything to the idea of selection based on relatedness is a somewhat contentious point.

The moral of this story is that in practice most, but not all forms of similarity selection take place as a result of relatedness or kinship.

Note that we are not talking about blood relatendess here. Organisms which are memetically related may also come to cooperate - as a result of the related memes manipulating their hosts.

Tuesday, 18 September 2012

Group selection and kin selection: formally equivalent

These days, there's a scientific consensus about group selection and kin selection.

As Marek Kohn said in 2008:

There is widespread agreement that group selection and kin selection — the post-1960s orthodoxy that identifies shared interests with shared genes — are formally equivalent.

As Michael Wade, 2009 put it:

It is remarkable that kin selection has been widely accepted and group selection widely disparaged when, for simple genetic models, they are actually equivalent mathematically.

As Peter Richerson, 2012 put it:

I think most evolutionists now agree that kin and group selection are the same thing.

Such observations date back to Hamilton (1975). Queller (1992) is another important paper on the topic.

Modern paper titles in the area include: "Group selection and kin selection: two concepts but one process" and "Group selection and kin selection: formally equivalent approaches".

In "Social semantics: how useful has group selection been?", West, Griffin and Gardner (2009) state:

There is no theoretical or empirical example of group selection that cannot be explained with kin selection.

The theoretical equivalence of kin selection models with those of the new group selection seems to be fairly widely recognized. Wilson and Wilson (2007) seem to agree, saying:

The theories that were originally regarded as alternatives, such that one might be right and another wrong, are now seen as equivalent in the sense that they all correctly predict what evolves in the total population. They differ, however, in how they partition selection into component vectors along the way. The frameworks are largely intertranslatable and broadly overlap in the kinds of traits and population structures that they consider.

Formal models of "group selection" and "kin selection" are now widely regarded as producing the same results. Gardner and Grafen (2008) say:

group selection has already been incorporated into social evolution theory, and is found to be exactly equivalent to kin selection: the two approaches are simply different ways of describing the same evolutionary process and both lead to the prediction that individuals should maximize their inclusive fitness

Here's Gardner, West and Wild (2011):

it has long been understood that the kin selection and multilevel (group) selection approaches to social evolution are mathematically equivalent, and merely represent different partitions of the same evolutionary process (i.e. natural selection; Hamilton, 1975; Grafen, 1984, 2006a; Wade, 1985; Frank, 1986, 1995; Queller, 1992b; Rousset, 2004; Gardner et al. , 2007; Lehmann et al., 2007b; Gardner & Grafen, 2009). No model of multilevel selection has ever delivered a (correct) prediction that could not be reformulated in terms of kin selection – despite repeated claims to the contrary.

Kerr and Godfrey-Smith (2002) recommend switching between the two perspectives - saying:

we also argue that each type of model can have heuristic advantages over the other. Indeed, it can be positively useful to engage in a kind of back-and-forth switching between two different perspectives on the evolutionary role of groups. So the position we defend is a “gestalt-switching pluralism.”

Group selection enthusiast Samir Okasha endorsed equivalence in a 2010 editiorial titled Altruism researchers must cooperate writing:

Lastly, kin and multi-level selection are not alternative theories; they simply offer different takes on the question of how social behaviour evolved. Proponents of kin selection, for example, explain sterile workers in insect colonies by saying that the workers are helping the queen to reproduce, and thus boosting their own inclusive fitness. Proponents of multi-level selection argue that the workers are providing a benefit to the colony as a whole, thus making the colony fitter than other colonies. These explanations may seem different, but mathematical models show that they are in fact equivalent.

To his credit, group selection enthusiast David Sloane Wilson is now on the correct side in this debate. So is group selection enthusiast Peter Richerson and group selection enthusiast Michael Wade.

For naysayers in this area, see here.


Group selection turns out to be kin selection

It was observed early on that most of the evidence that was cited as favouring group selection seemed to be explained rather well by kin selection.

Slime mold aggregation typically consisted of identical clones - the highest level of relatedness it is possible to get. The ant and termite families that group selection enthusiasts loved got a long list of nice quantitative predictions from kin selection theory. The evolution of chickens that supposedly counted as evidence for group selection was explained neatly by kin selection: the chickens were couped up with their close relatives.

It was argued that group selection was needed to explain cooperation between strangers who were not genetically related. However, this line of argument ignored cultural kin selection - much cooperation between strangers could be explained neatly in terms of shared memes - rather than shared genes.

There were other factors that explained cooperation too: including reciprocity, reputations, virtue signalling and the Allee effect. It seemed as though group selection was trying to take credit for the moves of other theories which were already well-established.

Was there anything group selection was needed for? Fortunately, this question now seems to have been answered...

The evolution of group selection

Group selection enthusiasts were dissatisfied with the consensus that group selection couldn't compete with individual-level selection. They counter-attacked in a number of ways, saying that:

  • The insistence by critics on groups being disjoint was mistaken - groups could still be targets of selection even if they overlapped.

  • Kin selection was actually a type of group selection - acting on family groups.
  • Reciprocal altruism was actually a type of group selection - involving small groups of reciprocators.
  • Some major evolutionary transitions showed group selection in action - today's organisms were once yesterday's groups.
  • Group selection had firm mathematical foundations - in the form of the Price equation.
  • Laboratory experiments on group selection showed that it was potentially a powerful force.

  • There are circumstances where between-group migration was very low - for example with parasites and symbiotes that spent most of their time inside other organisms.

  • Group selection explained altruism, religion, senescence, sexual recombination - etc.
They also pointed out that a number of criticisms that had been leveled at group selection were wrong:

  • The idea that between-group variation would be destroyed by migration and/or within-group selection was wrong - since substantial between group variation could be observed empirically - for example in a species such as our own.

  • The idea that groups are not replicators - and so aren't a valid "level of selection" - was bunk.
  • The widespread 50:50 sex ratio was not an argument against group selection for that trait - group selection could easily favour a 50:50 sex ratio.
A number of these points seemed quite reasonable - however...

Friday, 14 September 2012

Tag-Based Cooperation

Price's insight about the generality of selection made slow progress into the mainstream. Universal Darwinism is still struggling for acceptance in the mainstream, four decades after Price published. Another important aspect of the generality of selection - cultural kin selection - also laboured in obscurity for a long time. Some memetics enthusiasts understood it, but few others understood them. It wasn't until the year 2000 that the idea began to gather steam in academia - under the term "Tag-Based Cooperation".

A pioneering paper in 2001 - titled "Evolution of cooperation without reciprocity" - introduced the idea. The abstract said:

Here we use computer simulations to show that cooperation can arise when agents donate to others who are sufficiently similar to themselves in some arbitrary characteristic. Such a characteristic, or `tag', can be a marking, display, or other observable trait. Tag-based donation can lead to the emergence of cooperation among agents who have only rudimentary ability to detect environmental signals and, unlike models of direct or indirect reciprocity, no memory of past encounters is required.
"Tag-Based Cooperation" was recognised as a form of cultural kin selection by Sigmund and Nowak in 2001 - as follows:

the mechanism that leads to cooperation is a form of kin selection — either classical (if traits are inherited genetically) or social (if they are inherited culturally, like a dress code).


Tuesday, 11 September 2012

Enter George Price

One of the more significant developments in the early history of kin selection was the contribution of George Price. Price had read Hamilton's papers, and contributed a mathematical model of kin selection that proved superior to Hamilton. Hamilton had worked with the concept of shared genes. Price developed a largely-equivalent formulation based on correlations. Price's maths was better than Hamilton's - and it raised the possibility of spiteful behaviour.

Hamilton had predicted that close relatives would cooperate - since helping relatives helps copies of your genes inside them. Price pointed out that another way of helping your genes was killing off the least-closely related individuals in the population - that kin selection had an dark side: spite.

Spite was an interesting idea - but not a terribly practical one. The effort and risk involved in harming large numbers of non-relatives seemed highly unlikely to result in a payoff to the actor.

Price's equation allowed for a fairly simple derivation of Hamilton's rule. It was rather like the relationship between Newton's model and Einstein's. Price had a more general model that made almost all the same predictions as Hamilton's model - but handled a few extreme cases better.

However, Price's equation was to have other effects - besides its predictions about spite. Essentially Price modeled the effect of natural selection and persistence on the frequency of a trait in a population. It had two terms, one representing the persistence of the trait and the other representing the change caused by selection. It more-or-less ignored drift by dealing with expected values: stochastic forces would average out. Price had provided the logic of Universal Darwinism - and showed how the mathematics of selection could be applied to practically anything. Its generality reinforced the idea that blood relatedness was only one factor that could be responsible for trait correlations between generations. Many interpreted this to mean that the idea of "kin" was out - and that we needed a more general theory.

One of the things that the Price equation could be applied to was entire groups. You could split a population into groups (however you liked) and apply the Price equation to those groups. You could also apply the equation to individuals within those groups. That quickly led to a model of group selection. The change in frequency of a trait could be modeled as being divided into a component due to natural selection within groups, and a component due to natural selection between groups.

Price's maths could have led to a rapid reconciliation between kin selection and group selection proponents. However, that isn't quite what happened, as we will see.


Saturday, 1 September 2012

Kin selection and group selection: early rivals

The dust-up between kin selection and group selection began in 1964 with a paper by John Maynard Smith titled "Group Selection and Kin Selection". Maynard Smith wrote this after reading Hamilton's 1964 paper on the topic and delaying its publication. That caused some sour grapes - but we won't go into that story further here. The paper introduced and christened kin selection - and contrasted kin and group selection. Kin selection offered a promising explanation for altruism, while group selection seemed to have demanding prerequisites, and was probably not very significant. The paper established a pattern of thought on the topic that Maynard Smith would perpetuate throughout his career.

Hamilton wrote mostly about kin selection. Williams (1966) came out against group selection. Wilson (1975) promoted kin selection, and Dawkins (1976) was critical of group selection. Kin selection become popular, group selection became neglected and was believed to be an insignificant force.

Frank's 1998 book "Foundations of social evolution" barely mentions group selection. Alexander's "Darwinism and Human Affairs" (1979) says:

We can now say with conviction that Williams' argument against group selection was right.

Group selection was theoretically dubious, biologists seemed to rarely need to invoke it - and there wasn't any evidence for adaptations for the good of the group.

However, group selection refused to die quietly. Over the years its advocates became increasingly vocal in its defense. The "levels of selection" issue became one of the longest-running battles in evolutionary biology. The issue still seems to be alive and kicking in 2012.

On inclusive fitness

Inclusive fitness is an important concept in kin selection theory. It describes a concept which is an augmented version of "personal" fitness. Here is Hamilton's 1964 description:

Inclusive fitness may be imagined as the personal fitness which an individual actually expresses in its production of adult offspring as it becomes after it has been first stripped and then augmented in a certain way. It is stripped of all components which can be considered as due to the individual's social environment, leaving the fitness which he would express if not exposed to any of the harms or benefits of that environment. This quantity is then augmented by certain fractions of the quantities of harm and benefit which the individual himself causes to the fitnesses of his neighbors. The fractions in question are simply the coefficients of relationship appropriate to the neighbors whom he affects; unit for clonal individuals, one-half for sibs, one-quarter for half-sibs, one-eighth for cousins [...] and finally zero for all neighbors whose relationship can be considered negligibly small.

Inclusive fitness can be seen as an attempt to save the concept of organism-level fitness in the aftermath of the development of the idea of kin selection. The concept and its usefulness can be rather challenging to understand.

The term "inclusive fitness theory" has become a rival to the terminology used by kin selection theorists.

The term "inclusive fitness theory" was endorsed by Hamilton - the originator of the idea of kin selection. Hamilton (1975) wrote:

The usefulness of the 'inclusive fitness' approach to social behaviour [...] is that it is more general than the 'group selection', 'kin selection', or 'reciprocal altruism' approaches and so provides an overview even where regression coefficients and fitness effects are not easy to estimate or specify.

the foregoing discussion shows that kinship should be considered just one way of getting positive regression of genotype in the recipient, and that it is this positive regression that is vitally necessary for altruism. Thus the inclusive-fitness concept is more general than ‘kin selection’.

However, here we will use the term 'kin selection' in perference to 'inclusive fitness theory' - and treat them as synonyms.

Much is made by advocates of using "inclusive fitness" terminology of the virtues of identity - as opposed to "identity by descent". It is true that it is identity that matters in kin selection. However, in biology, most identity is identity by descent. The alternatives to identity by descent are identity by chance and identity by convergent evolution. However for complex biological objects, identity by chance would be a miracle, and identity by convergent evolution depends on convergent evolution being a very powerful and specific force - which it mostly isn't.

In the realm of organic evolution, genetic identity is identity by descent. If two stretches of DNA of a non-trivial size are identical, it is because they are descended from a common ancestor.

In cultural evolution, identity by convergent evolution is more likely - but still so rare as to be a big non-issue. Convergent evolution usually produces similarity - not identity.

The term "kin selection" is short and neat. It fits in neatly with the term "group selection". "Inclusive fitness theory" is a mouthful. It refers to an esoteric technical concept. It makes no mention of kinship or family.

In a nutshell, the terminology of "inclusive fitness theory" totally sucks.

An introduction

The term "kin selection" refers to a type of selection in which heritable elements increase in frequency by causing their bearers assist their relatives.

Genes coding for strategies favouring assisting relatives can spread in populations if the benefit to the relatives multiplied by a coefficient of relatedness is greater than the cost to the actor. This idea is formalised in Hamilton's equation: rB > C.

The theory of kin selection successfully models much of the apparently-altruistic behaviour in the world. It explains the love of mothers for their children. It shows that an individual-centric perspective on the natural world is incorrect - organisms are built to care about others besides themselves.

The theory of kin selection has roots dating back to Darwin, but it was first formalised by Hamilton in 1964. George Price contributed a slightly improved formalisation in the 1970s - which showed how kin selection could favour spite - as well as altruistic behavior.

Kin selection has been a tremendously productive theory. However in modern times much mud slinging against kin selection has taken place by advocates of a supposedly alternative theory of the evolution of social behaviour: group selection. Here we will attempt to set the record straight on the topic.