Daniel Dennett: Information, Evolution, and Intelligent Design

Cultural evolution starts 7 minutes in. Memes start 18 minutes in.

48 minutes in Dennett has a strange section about how internet memes are a contradiction in terms (because they are designed rather than evolved). Few meme enthusiasts would agree with Dennett here, I think.

Acerbi and Mesoudi on the genotype/phenotype split

Acerbi and Mesoudi discuss the genotype/phenotype split in a recent paper [*].

The picture they criticize is the picture I prefer - and I will respond to their critique. First, here is what they say:

One proposed solution to this puzzle is to consider the information, wherever stored, as the equivalent of the biological genotype, and the expression of the information in behaviours or artifacts as the equivalent of the biological phenotype (Dawkins 1976). The problem here is that it assumes that, when copying, we have access to a “cultural core” (Sperber and Claidière 2008), which represents the information/genotype, which we then use to build variable phenotypic expressions. This might be loosely the case: the classic example is the transmission of a recipe to cook, say, lasagne, where the recipe represents the transmitted, stable, genotype, and what you serve to your guests at dinner is the variable phenotype. However, in many cases, we do not have access to a “recipe”, but we extract the information from the result/phenotype (such as when we try to reproduce lasagne after tasting it at a friend’s home). Richerson and Boyd (2005) make a similar point when noting how the mental representations of different individuals who have tied the same bowline knot might in principle be very different. What is the genotype here? The individual, variable, mental representations of the bowline knot cannot be the genotype, as they are not, in general, transmitted, because they are different. For the same reason, the information stored in the artifact itself does not transfer directly in the (variable) mental representations.

In response I would deny the assumption at the end of the second sentence. The picture of genotype-as-inherited information and phenotype as genotype products does not assume that there's a "cultural core" - a recipe to be copied. Some of the "genotype" can go on to reside in the cake. The rule here is that it is germ-line if it is copied from.

I agree that this picture of the genotype/phenotype split is imperfect. However, the genotype/phenotype split is a very productive concept in many cultural and organic domains. IMO, putting the split between genotype-as-inherited information and phenotype as genotype products is the best place to put the divide. It works well in both organic and cultural evolution.

The rule that it is germ-line if it is copied from also seems fairly neat to me - by contrast with the idea that it is germ-line if it is the product of copying.

Blackmore (1999) wrote:

I will not, therefore, use the concept of the meme-phenotype because I cannot give it a clear and unambiguous meaning.

I claim that the picture of genotype-as-inherited information and phenotype as genotype products is both clear and unambiguous. It also happens to be the standard meaning of the concept - as far as I can tell.

IMO, the phenotype-genotype split is essential to both organic and cultural evolution. Trying to skip the distinction is a feeble cop-out. The issue is not whether to make a divide, but where to make it. Criticism of one proposal needs to be accompanied by a better proposal - if it is to cut any ice. This is where Acerbi and Mesoud don't really come through. They can see flaws in the proposal that I prefer - but they don't really have anything better to offer. Consequently, I think their criticism fails. Imperfections are not enough - critics need to have a better proposal.

Here's what they say next:

While this may appear pessimistic, we believe that pluralism in the conceptual definitions of the unit of analysis in cultural evolution is not a problem (see also Lymann and O’Brien 2003; O’Brien et al. 2010). Biologists, too, work simultaneously with multiple concepts of the ‘gene’, varying with context and use (Stotz and Griffiths 2004). Depending on various domains, and on the questions one is interested in, an opportunistic strategy can be the best choice.

Pluralism. Maybe. Of course the problem with pluralism is confusion and ambiguity. You can have a plurality of concepts without them clashing over terminology.

In the case of their example, 'gene' the main problem is the usage by molecular biologists. The term 'gene' belongs to evolutionary biology. The molecular biologists should simply abandon their claim on the word.

References

• Acerbi and Mesoudi (2015) If we are all cultural Darwinians what’s the fuss about? Clarifying recent disagreements in the field of cultural evolution

What does it mean to say that memes consist of information?

It seems to me that there's a lot of confusion surrounding the idea that memes consist of information. To give some examples, Here's Ted Cloak, apparently advocating a kind of behaviourist memetics:

You may note that I don’t think memetics concerns information, or ideas, or mind, or thought, or consciousness, or language. I assert that culture, and therefore memetics, is simply about behavior. Its job is to explain why people and some infrahuman animals do certain things.

...and here's John Wilkins:

For some time now [1] I have had problems with the notion of information. Not, please note, with this or that piece of information, but with the notion itself, especially in the natural sciences. In this age of computers and internets, we have taken to mistaking the thing described for the thing itself, and treat information as a property out there in the world, not a representation in our heads and language.

My perception is that a lot of these kinds of objection come from not knowing enough about information theory.

Anyway, perhaps a few clarifications are in order. I generally endorse and promote the standard Shannon/Weaver notion of information - that the concept of "information" makes sense in the context of an observer and it represents the aspects of a message that they don't already know.

However, many have struggled to see how to relate this sort of information to evolutionary theory - where it isn't obvious what observer we should be considering. Having a scientific quantity being observer-dependent seems to limit its usefulness. Surely there will be arguments about measurements by different observers. Scientific consensus and objectivity will surely suffer.

There are other concepts competing for the term 'information' (for example, Fisher information). Often these are not observer-dependent. Why should scientists use an observer-dependent quantity?

I think that Shannon/Weaver information provides the most popular meaning of the term "information" - and that its popularity is well deserved. As for scientific objectivity, I think the concept of a "reference observer" is useful. To avoid ambiguity, scientists can specify the observer. For example they might specify an observer that knows what A, T, G and C base pairs are - but has no knowledge of their likely sequence. They consequently assume the maximum entropy distribution - in which all sequences are equally likely.

Yes, it's possible to debate which "reference observer" is most appropriate - but in practice there are relatively few such debates. Much the same problem afflicts the notion of "complexity" - but that is still a useful concept.

Once you have an observer then the issue of what counts as information to that observer becomes simple - it's anything they don't already know. If you hypothesize a highly ignorant reference observer, then practically any possible message qualifies.

Another perspective on information comes from dimensional analysis. Many scientific quantities have attached dimensions: units of time distance, mass - and so on. Information is dimensionless. However, you can measure quantities of information - in 'bits'.

Information is what can be transmitted over the internet. It is measured in bits. Information is portable. It is substrate neutral. The same message can be transmitted in a variety of different media.

Saying that memes carry culture or that they are inherited are making fairly specific statements about what counts as a meme and what does not. Saying that memes are made of information is also a kind of constraint - but it's a kind of negative constraint, an avoidance of more constraining constraints. It conflicts with the idea that memes are necessarily neural structures, or that they behaviours or artifacts. The picture of memes as information allows memes to exist inside brains, behaviours and artifacts. Their inheritance pathway can involve a type of biological metamorphosis - as they move from one substrate to another.

Some apparently think that there's more to cultural transmission that inherited information - and thus more to cultural transmission than memes. The internet illustrates that you can learn almost anything from pure informational sources. However it is possible to point to some types of martial arts, massage techniques as being very challenging to learn over the internet. Rather that indicating some kind of non-informational cultural transmission, it seems to make more sense to me to consider this as a current limitation of our recording and playback facilities - one that is likely to be rectified in the future.

In my humble opinion, this picture of memes as inherited cultural information aligns neatly with the picture of genes as inherited information from evolutionary biology. It allows information theory to act as a common backbone for memetics and genetics. I go into this in my informational genetics article.

Donald Campbell: Pioneer of Universal Darwinism

It's well known that Donald Campbell was an early pioneer of cultural evolution. Most in the field acknowledge his work as influential or historically important. One of his early papers on "Blind variation and selective retention" dates from 1960. Later in life, Donald Campbell explored the limits of evolutionary theory - applying it to a range of inorganic phenomena. In particular, there's the following paper:

Bickhard, Mark H. and Campbell, Donald T. (2003) Variations in Variation and Selection: The Ubiquity of the Variation-and-Selective-Retention Ratchet in Emergent Organizational Complexity. Foundations of Science, 8(3), 215–282.

Here, Campbell and Bickhard apply the principles of variation and selection to a range of phenomena - including why gravel accumulates at the edges of roads, crystal growth, crystal stability, the formation of atoms and molecules and catalysis. They argue that "energy wells" are common causes of selection phenomena - giving rocks, planets, stars, and galaxies as examples.

Campbell mostly avoids the terminological debate about whether such phenomena qualify as being "Darwinian" - which is still the source of much modern noise and confusion. Instead he uses terminology oriented around the terms "variation" and "selection" and examines whether they produce "goodness of fit", "adaptation" and "evolutionary historicity".

There seem to be some differences between my understanding of this area and Campbell's (which admittedly dates from the 1990s). I would give examples of tree-like phenomena in nature - such as electrical discharges, propagating cracks, fractal drainage patterns and diffusion-limited aggregation - and say they these trees are family trees, showing clear ancestor-descendant relationships. I make that argument in more detail on my positional inheritance page.

Another apparent oddity of Campbell's perspective is that none of his examples of selection seem to involve things being destroyed. Death is a common source of selection in the organic realm - so it seems natural to me to give instances of destruction as examples of selection in the inorganic realm. For example, the destruction of islands by the sea, of rocks by landslides, of sediments by subduction, and of pebbles by erosion all seem to me to be fine examples of selection. Campbell seems to me to systematically steer clear of destructive examples. However, this seems like a very curious thing to do. Did Campbell really not regard destruction as being a source of selection?

Though Campbell still seems to me to have a bit of a restricted perspective on the scope of selective explanations, he applied the concepts of selection, fitness and adaptation deeply into the realms of physics and chemistry in the 1990s. That makes him into one of the pioneers of Universal Darwinism.

Brent Jesiek's history of memetics

In 2013 I called for a history of cultural evolution. Something like a more fleshed-out version of my own Memetic Timeline.

However at that stage I hadn't seen Brent Jesiek's 110 page MSc thesis. This presents a comprehensive history of memetics. It mostly concentrates on the period from 1975 to 2003. It is available free of charge online - to ResearchGate members.

• Jesiek, Brent K. (2003) Betwixt the Popular and Academic: The Histories and Origins of Memetics.

Brent Jesiek's history is comprehensive and impressive. It's a history of memetics - rather than a history of cultural evolution - focusing heavily on those thinkers that dealt with the possibility of there being cultural equivalents of genes. This rules out much of the work done in academia on cultural evolution - much of which is still very confused and muddled about this point.

Unfortunately, this focus leaves out much of interest - and some of what it puts in its place is not too interesting. For example, there's quite a large section devoted to the efforts of Aaron Lynch. Alas, Lynch's book on memetics was pretty terrible. Paul Marsden's account of how bad it was is of much better quality. Also, Susan Blackomre doesn't get much space in this history - which doesn't seem very fair, given the scale of her efforts.

Anyway, despite some misplaced emphasis, Brent Jesiek's history is an essential guide to the history of memetics. It's great to have such a resource available online.

Cultural evolution vs history

Some wondered why we call it "cultural evolution" - rather than "history".

In 2012, Steven Pinker asked:

Does Cultural Group Selection Add Anything to Conventional History?

In 2015, Richard Lewontin asked:

Why do you use cultural evolution instead of cultural history? Why evolution instead of history?

To me these are odd questions to ask - but I think there are reasonable answers:

• The term "history" has traditionally referred to human evolution after the invention of writing. By contrast, cultural evolution goes back many millions of years and also applies to non-human animals.

• The term "evolution" conjures up Darwin's famous explanation of how evolution operates. The term "history" fails to do this. The association is appropriate.

• History has traditionally been studied as part of the humanities. The humanities have historically been characterized by poor quality scientific traditions. In particular, historians widely rejected theory, picturing theories as preconceptions which could distort the facts. As a result, history increasingly turned into a fact-gathering exercise. This is, of course, not a scientific approach to the topic. As a result, many scientists don't want to associate themselves with historians. The historians dirtied their own nest, and many scientists don't want to be tainted by their stench.

We have the terms "evolution" as well as "natural history". They don't mean exactly the same thing. "Evolution" traditionally refers to change - whereas "history" can cover both change and stasis. Also, "evolution" has stronger connotations of gradual change.

Cultural evolution: evolutionary frontier

I've talked in the past about Cultural evolution's scientific lag. At first glance it might seem as though cultural evolution is a scientific backwater. There are few conferences or journals. Representation at universities is very patchy. Funding is poor. Sympathetic colleagues are hard to find and progress has been depressingly slow. This doesn't seem like a very attractive package to a budding young scientist. So: what is the attraction?

Though in one sense it is true that theories of cultural evolution lag behind their organic counterparts, in another respect cultural evolution is on the leading edge of evolution itself. If you look at most recent significant changes in the world, many of them involve cultural evolution. For example, memes - more than genes - are responsible for space travel, computers and the internet. Cultural evolution is on the cutting edge.

Cultural evolution is also on the leading edge of evolutionary theory. Organic evolution is a done deal - and has been for over a hundred years. There, researchers are mostly putting the icing on an existing cake. Cultural evolution is where the real action is. It is where new researchers can make an impact and make important discoveries.

Cultural evolution is of enormous social and political significance. A proper scientific understanding of how culture evolves is critical for making good policy decisions. Cultural evolution is much too important to be left to cultural anthropologists, who have failed to get to grips with the topic for over a hundred years and seem to suffer from poor scientific literacy.

Lastly, the role of cultural evolution looks set to become ever more important as time passes. In particular, memetic algorithms - which emulate cultural evolution - look set to play a critical role in the development of machine intelligence. Memetic algorithms and memetic programming are similar to genetic algorithms and genetic programming - only they are inspired by cultural evolution rather than organic evolution. Machines, like their human inventors before them, look set to harness the power of cultural coevolution - in order to attain the rapid rate of change which will fuel their future expansion and prosperity. The "code rush" as some call it. It is on.

Tim Lewens: Cultural Evolution: Conceptual Challenges

Tim Lewens has a book on cultural evolution coming out later this year: Cultural Evolution: Conceptual Challenges. For Google's preview of the book, see here.

His 2004 book - Organisms and Artifacts: Design in Nature and Elsewhere - also related to the topic.

I have generally dismissed Tim Lewens in the past as a feeble-minded meme critic who doesn't know what he is talking about. However the blurb to this book weakly suggests that he is in the process coming round to a sympathetic understanding of cultural evolution. Or maybe not, we will see. Here is the blurb:

Tim Lewens aims to understand what it means to take an evolutionary approach to cultural change, and why it is that this approach is often treated with suspicion. Convinced of the exceptional power of natural selection, many thinkers - typically working in biological anthropology, cognitive psychology, and evolutionary biology - have suggested it should be freed from the confines of biology, and applied to cultural change in humans and other animals. At the same time, others - typically with backgrounds in disciplines like social anthropology and history - have been just as vocal in dismissing the evolutionary approach to culture. What drives these disputes over Darwinism in the social sciences?

While making a case for the value of evolutionary thinking for students of culture, Lewens shows why the concerns of sceptics should not dismissed as mere prejudice, confusion, or ignorance. Indeed, confusions about what evolutionary approaches entail are propagated by their proponents, as well as by their detractors. By taking seriously the problems faced by these approaches to culture, Lewens shows how such approaches can be better formulated, where their most significant limitations lie, and how the tools of cultural evolutionary thinking might become more widely accepted.

---

Update 2016-02-08: C. Heyes review. From this polite review, the book doesn't sound to me as though it is going to be much good.

Here's a 2011 critical essay, showing Tim's perspective.

Sylvain Magne: What is a meme?

Here's Sylvain Magne - with a 20 minute presentation on the definition of the term "meme".

There's a transcript here.

I'm pleased to see discussion of the topic. However, since space and time are limited, I'll mostly confine my comments to the points where there is disagreement.

I think this topic is best explored using the infrastructure and terminology of information theory. Information theory has useful concepts that formalize this topic - such as the idea of Shannon mutual information - which is useful for formalizes the notion of copying. This article suffers from failing to build on this previous work.

Sylvain defends the notion of a "replicator" - which has proved to be a controversial term. The concept of as replicator was originally promoted by Richard Dawkins - with the admirable aim of enlarging the scope of evolutionary theory beyond the realm of DNA genes. However it has also resulted in much misunderstanding, confusion and criticism. Though for many, it's a foundational concept for memetics, I've generally been quite critical of the replicator terminology.

The biggest problem is that the etymology of "replicator" implies high-fidelity copying - whereas most models of evolutionary processes accept the copying fidelity as a parameter - and do not insist that copying be high fidelity.

I think that the best way to defend the notion of a "replicator" is to abandon the notion of high fidelity copying. That's the approach I take with "repology". This makes "replicator" into a misnomer - but this is still the best option for those wanting to keep the terminology.

Sylvain presents a defense of the "replicator" concept that preserves its notion of high-fidelity copying. His defense hinges on the concept of a "reader". Sylvain's "reader" is a system which identified whether two copies are identical or not. Sylvain gives the example of key copying to illustrate the concept. The lock acts as a reader and determines whether keys are functionally identical or not. Certainly in many evolving systems there are "readers". These typically perform error correction and detection. DNA copying features physical systems which act as readers. The same is true for must cultural systems which copy words. However, for other systems, it is not obvious that a "reader" exists. When ants copy each others' pheremone trails, there's no system deciding whether the behaviours are identical copies or not. Nature often doesn't care much about whether copies are identical or not. It sometimes cares about similarity - but that's a bit different. Rather than dividing the world of copies into those that are identical and those that are not, it is usually better to consider identity to be the extreme end of a continuum of varying levels of similarity.

Scientists sometimes care whether two systems are identical or not. However nature doesn't insist on the critera they use - and different scientists may use different criteria. A geneticist might treat genes with the same base pair sequences as being identical - while someone studying proteins might have a different idea about what 'identical' means in this case.

The concepts of "replicators" and "readers" may seem attractive when dealing with digital genes and memes - but they seem more like added complication when dealing with more general versions of evolutionary theory - where high-fidelity copying is not necessarily present. There, the concept of imperfect copying seems simpler. Variable-fidelity copying makes the concept of a "replicator" functionally redundant. The concept of "similarity" is broader and richer than the idea that copies are either identical or they are not.

Cultural infant mortality

In the organic realm, infant mortality is an important observed phenomenon, with an elevated infant mortality rate being observed in a wide range of species. The study of elevated infant mortality from an evolutionary perspective is part of what is known as Life history theory.

Several factors account for elevated infant mortality - for example:

• The small size of infants makes them less able to store resources - and thus more vulnerable to resource fluctuations.
• Some infants are widely dispersed but face patchy environments - where not all of them can thrive.
• Infants are often produced in huge numbers - and there aren't enough resources available for them all to survive to adulthood.

In the 1930s, Ronald Fisher proposed a general concept that covers many of these ideas - known as reproductive value. Reproductive value varies over the lifespan of an organism, reflecting their future reproductive potential. Old organisms have low reproductive value (their expected lifespan is lower and their fertility is reduced) and often infants do as well - due to the kinds of factors mentioned above. Fisher proposed that mortality rates could reasonably be expected to be optimised by natural selection to be proportional to the inverse of reproductive value.

Reproductive value is a concept which is closely related to fitness. Like fitness is is quite a general concept. However, as with fitness it is worth distinguishing between actual reproductive value (measured after the fact) and expected reproductive value - which is calculated on the basis of some other predictive theory about how inherited traits and the environment combine to affect the performance of the organism.

Though the concept of reproductive value is very general it is also often vague. For example, it follows that if infants are produced in huge numbers in each generation they will outstrip their resources, their reproductive value will be low - and there will be high levels of infant mortality. However here, high infant mortality follows directly from high birth rates - and invoking the concept of reproductive value didn't really help very much. It also doesn't help to answer the question of why so many offspring are produced in the first place. Isn't mass infant death very wasteful? In the case of widely dispersed seeds facing a patchy environment, some waste seems inevitable. However, in other cases, natural selection between the offspring may play an important role - weeding out those organisms with deleterious mutations or bad gene combinations by making sure that they "fail fast".

This brings us to cultural infant mortality. Culture provides a new domain for life history theorists with many interesting examples. What can be learned? What can life history theory contribute?

First there are some similarities. As with DNA genes, some memes face a patchy environment. Most flyers are trampled into the ground; the street preacher's ranting mostly gets no further than the sidewalk - and so on. Memes are also mass produced in far greater numbers than can ever survive. Radio and TV signals are broadcast in all directions. Some make it into space, where they could survive for a long time, while most others quickly hit dirt and turn into heat. Start-up companies and IT projects also exhibit high rates of infant mortality.

Also as with genes, many memes are end-of-line copies - with a limited lifespan and a low chance of personal reproduction. Most artifacts are like this. They are the equivalent of somatic cells of cultural evolution - their primary purpose is to assist the reproductive memes in the factory that made them - via cultural kin selection. Life history theory treats these end-of-line copies a bit different from germ-line copies.

Unlike DNA genes, the memes in artifacts often don't have very flexible control over their associated life history variables. If your main strategy for persisting is to be hard and strong that doesn't result in very much flexibility regarding senescence rates. DNA has mastered regeneration - and can more flexibly allocate resources to maintenance processes over the course of a lifespan. Artifact regeneration is a thing - as some automobile owners can attest. However, many artifacts are hard for end users to repair and they often get trashed at the end of their natural lifespan.

Another thing that most memes are not very good at yet is growth. Without growth, infants are not small, and so suffer less from predation and mechanical insults. Of course, some cultural forms do grow. Cities, roads networks and telecommunication networks all grow. However, most artifacts don't really grow - and without growth there is much less scope for infant mortality. Many memes aren't good at growing today. However we are still near to the origin of cultural evolution and it seems reasonable to expect that this limitation will disappear once we have easy access to robust molecular manufacturing technology.

High infant mortality is often regarded as a bad thing. However from the perspective of Darwinian processes, high infant mortality has some desirable aspects. If something is going to fail it is often best if it fails fast. Investments in components that are going to fail are often bad investments: it is better to spend the resources on something that is not going to fail. For many long-lived organisms there's a high-intensity selection process around the time of conception: gamete selection. More failures can occur during gestation and around birth. Rather than lamenting these failures, Darwinism suggests that we should regard them more as part of a natural process of weeding out the weak and unfit before they can do more serious damage to a family's resources.