Sunday, 17 March 2013

Nature's Oracle is out

Nature's Oracle is now available on Amazon.

A review: Nature's Oracle: The Life and Work of WD Hamilton by Ullica Segerstrale.

Razib Khan weighs in on the topic.

Robert Boyd critiques kin selection

This is a pretty technical presentation by group selection advocate Rob Boyd on kin selection and the dust-up relating to it in Nature. It's slides-and-audio only (sorry).

Boyd calls Stuart West and Andy Gardner the "Inclusive fitness mafia" 31 minutes in.

He presents an objection to kin selection theory, calling Hamilton's rule "true but misleading" 50 minutes in. It seems like sour grapes to me.

Tim Tyler: Wilson, The Social Conquest of Earth (review)


Hi. I'm Tim Tyler, and this is a review of this book:

The Social Conquest of Earth by Edward O. Wilson
The book is about the rise of social organisms on the planet and how they have come to dominate their ecosystems. Wilson uses ants and humans as his main examples of the success of sociality.

It's aiming to be a popular book. There are, in fact, a few scientific references in the book, but if you're not careful, you won't find them until you get to the very end - where they are clustered together in a list.

The first 100 pages of the book is a history of the rise of humans. Alas, I'm one of those who often finds histories boring, so I had to push a bit through this section of the book. Then there's thirty pages on social insects. Then there's sixty pages on the theories of social evolution. This is really where Wilson comes unstuck.

Wilson launches into a misguided attack on kin selection theory. Obviously, Wilson has some understanding of kin selection. However, he doesn't have a sympathetic understanding of it, and instead seeks to destroy and discredit it. Alas, in order to do this, he has to set up a series of straw men and then vigorously attack them. This might be entertaining for some - but Wilson is a popular author, and here we run the risk of an eminent scientist confusing and misleading the next generation of students. About the best that can be said for Wilson's attack on kin selection is that it is embarrassing and half-baked. I noticed that, as if to illustrate that he doesn't really know what he is talking about, Wilson gives an incorrect definition of inclusive fitness, saying:

The inclusive fitness of the individual is its personal fitness, in other words the number of its personal offspring who grow up and have children of their own added to the effect its actions will have on the fitness of its collateral relatives, such as siblings, aunts, uncles, and cousins.

That definition is wrong. Those familiar with the topic will recognise that it's only two thirds of the correct definition. Wilson goes on to discuss the claimed equivalence of kin and group selection. He says:

Further there are mathematical difficulties with the definition of r, the degree of relatedness. These difficulties render incorrect the oft-repeated claim that group selection is the same as kin selection expressed through inclusive fitness.
After supposedly demolishing the decades of work by kin selection theorists, Wilson goes on to present his preferred alternative theory of social evolution - which is based on group selection and the idea of group-forming forces - where Wilson gives a "defensible nest" as his main example of the latter.

Unfortunately, Wilson's book is probably exhibit number one when it comes to showing how group selection leads to confusion about evolutionary theory. Wilson is evidently in a hopeless muddle about kin selection and group selection. It seems hard to avoid the conclusion that group selection lies at the root of the problem. While Wilson might have been trying to help the cause of group selection in his book, it seems more likely that it will be an embarrassment for group selection's proponents.

Group selection's main problem that it faces a well-armed opponent. Kin selection explains all the same things that group selection explains - and has been scientific orthodoxy for decades. Most modern group selection enthusiasts seem to have finally agreed that group selection makes no predictions not also made by kin selection. The main case for group selection these days seems to amount to the idea that it represents a stimulating alternative perspective. High-level selection can be a form of holism to counter kin selection's reductionism. Group selection models can be tractable in some cases where lower-level kin selection models are not. This is all reasonable: group selection does have a positive side. The problem with group selection is that historically its use has often led to junk science. People tend to invoke benefits to group willy-nilly - without doing the math, or even sanity-checking their explanations. To many scientists, group selection explanations thus often look like a poor substitute for actually understanding what is going on. Many explanations based on group selection, on closer examination, have turned out to be wrong or misleading. In the cases group selection is useful, kin selection is often more useful. SO, group selection has its merits - but needs a health warning.

Wilson claims that humans are a eusocial species - even though we mostly lack one of the primary traits associated with eusociality - namely reproductive division of labour. Wilson doesn't explain that his use of the term is unorthodox and controversial.

Wilson's examples of eusociality are all from animals. He lists beetles, aphids, thrips, shrimps, mole rats and humans. However there are other examples of social behaviour among microbes, plants and fungi. In particular multicellularity is a common form of eusociality. The examples of its evolution illuminate the topic of social evolution. However, Wilson makes no use of these examples.

Wilson attempts to apply the study of animal social evolution to humans - and this leads him to the kin selection (or group selection) axis. While important, this leads to a pretty myopic perspective on human social behaviour, which is heavily influenced by reciprocity, reputations, byproduct mutualism - and manipulation of humans by memes. You can't really coherently discuss human social evolution without casting your net wider than this.

Lastly, the book ends up with a review of cultural evolution in humans. In the 1980s Wilson was a pioneer in applying biology to human culturally-transmitted behaviour, writing two books on this topic. Wilson at least understands that culture evolves - and that memes coevolve with genes. However, his older books promoted a pre-memetics perspective, in which everything boiled down to DNA genes. Checking to see if Wilson had moved on 40 years later, I was surprised to find that little had changed. He cited his old work, and trotted out much the same theory that he advocated back in the 1980s. This idea has been totally superseded in the mean time by the application of Darwinian evolutionary theory directly to cultural variation - along the lines pioneered by Richard Dawkins in 1976. Wilson's discussion of this work is confined to one paragraph. He chastises the researchers involved for paying insufficient attention to the interactions between genes and memes - and then launches into a laundry list of such interactions.

This hopeless treatment of cultural evolution might not matter - but this is one of the few books addressing the topic that I regularly see in bookshops. This might be the first exposure to the topic for many. A teaching opportunity has been squandered, because Wilson is stuck back in the 1970s on the topic and has failed to grok most of the more recent literature.

In the book, Wilson expands on his old claim that genes keep memes on a leash by presenting an argument against the possibility of a memetic takeover. He says:

I am further inclined to discount the widespread belief that robotic intelligence will overtake and potentially replace human intelligence.

What are Wilson's reasons? He goes on to give them, saying:

To advance from robot to human would be a task of immense technological difficulty. But why should we even wish to try? Even after our machines far exceed our outer mental capacities, they will not have anything resembling human minds. In any case, we do not need such robots, and we will not want them. The biological human mind is our province.

Wilson's objection seems simple-minded to me. He gives me no reason to think that he knows much about machine intelligence or futurism. He's just a famous biologist speculating on something about which he knows little. The idea that people will lack motivation to build superhuman intelligent machines just seems daft to me. The motivation is obvious.

So: this is not a good book. It is quite readable, but it's full of outdated or wrong science. Readers should be warned that, while Wilson speaks authoritatively, he doesn't really know what he's talking about in many places.


Saturday, 9 March 2013

Tim Tyler: Alexander, Darwinism and Human Affairs (review)


Hi. I'm Tim Tyler, and this is a review of this book:

Darwinism and Human Affairs by Richard Alexander
The book was published in 1979. The contents represent what we would now call sociobiology. Unfortunately, this is the bad kind of sociobiology that tries to trace everything back to DNA genes.

The book offers a perspective similar to that of E.O. Wilson - who published similar works at around the same time.

I read it mainly in order to understand Richard's perspective on human culture from that era.

The book predates our modern understanding of cultural evolution and instead presents a view of culture based on DNA genes.

In a section titled: A comparison of organic and cultural evolution, Richard breaks cultural evolution down into inheritance, mutation, selection, drift and isolation. He then reviews existing literature relating to the modern perspective on cultural evolution citing the views of Dawkins, Durham, Cloak, Cavalii-Sforza, Feldman, Richerson and Boyd on the topic. This might all sound good. However, the section then goes on to dismiss this material, saying:

regularity of learning situations or environmental consistency is the link between genetic instructions and cultural instructions which makes the latter not a replicator at all, but in historical terms a vehicle of the genetic replicators.
Having thus dismissed the Darwinian view of culture which does so much useful work in modern times, Richard offers in its place the idea that cultural information is a persistent part of the environment that is influenced and manipulated by genes. This is true - as far as it goes. However, these days, we know that you can get a lot of mileage out of the idea that cultural variation evolves in a very similar way to the way genetic variation evolves - complete with selection, adapatation, drift, recombination, kin selection, gradualism and heritage constraint. Richard's model is not exactly wrong, it just isn't as helpful as it could be. Because culture evolves, evolutionary theory has a rich array of tools and models which can be directly applied to cultural variation without much modification. If you fail to recognise that, you don't get to use these tools and models.

It is rather frustrating for the reader to see the author toy with the correct theory - and then abandon it. On a more positive note, it was probably good to have the position that everything boils down to genes clearly articulated - to give critics something to argue against.

Most of the rest of the book is concerned with explaining various types of human behavior in terms of DNA genes. Richard looks at nepotism, incest avoidance, xenophobia and various other aspects of human affair with a biological basis. He engages with anthropologists critical of biological approaches. While he makes most of them look stupid, some of their objections make some sense retrospectively. One thing they were critical of was the over-application of kin selection theory. Looking at the kin selection being advocated back then, only kinship between DNA genes was used - and all other kin-like relationships were classified as fictive kinships, with the organisms involved being manipulated and fooled into thinking their colleagues are their kin. This now looks like a terribly impoverished kind of kin selection, in the light of cultural kin selection. There's memetic kinship, as well as genetic kinship. The anthropologists didn't have the concept of cultural kin selection back then either, but at least they recognized that there was much more to kinship than relatedness between DNA genes.

Richard bashes group selection here and there in the book, but his story of how humans evolved revolves heavily around conflict between groups. Because of this, modern group selection enthusiasts may find him an ally as much as an adversary. Here's a sound bite from the book on the non-equivalence of kin and group selection:

Despite the recent prominence accorded to the view that the maximization of inclusive fitness by helping the aggregate of one's relatives - or what Maynard Smith called "kin selection" is a kind of group selection, this is a misleading, if not erroneous view. Group selection thwarts the reproductive interests of individuals when these interests differ from those of the group. Kin selection is a way in which individuals further their genetic interests via other individuals who carry some of their genes.

While this might have sounded plausible at the time, it now looks wrong. Group selection has turned out to be a way in which heritable information perpetuates itself via promoting copies of itself in other group members.

The book finishes up with a look at justice and ethics. Richard says:

Arguments given above, and the cited references make it clear why I believe that evolution has more to say about why people do what they do than any other theory. In contrast, my answer to the question "What does evolution have to say about normative ethics, or defining what people ought to be doing?" is "Nothing whatsoever."

I think this is a cop-out. Evolutionary theory surely has more to say about this subject than any other branch of science. If intelligent people aren't supposed to use science to guide their ethics, then it isn't clear where they are supposed to be getting their ethical principles from. Richard says that he thinks that pain and pleasure are the basis of most normative ethics - but even if one accepts that, classifying these as "non-evolutionary" ignores the idea that the brain evolved, and the idea that the brain evolves over an individual's lifetime, using copying, variation, and selection in order to to seek its rewards. I don't think you can get away from evolutionary ethics this easily.

This is a nice book to read, but pretty out-of-date these days. However it is interesting to see part of the background that modern theories of culture and human evolution were developed against.