Temes: An Emerging Third Replicator

All around us information seems to be multiplying at an ever increasing pace. New books are published, new designs for toasters and i-gadgets appear, new music is composed or synthesised and, perhaps above all, new content is uploaded into cyberspace. This is rather strange. We know that matter and energy cannot increase but apparently information can.

It is perhaps rather obvious to attribute this to the evolutionary algorithm or Darwinian process, as I will do, but I wish to emphasise one part of this process — copying. The reason information can increase like this is that, if the necessary raw materials are available, copying creates more information. Of course it is not new information, but if the copies vary (which they will if only by virtue of copying errors), and if not all variants survive to be copied again (which is inevitable given limited resources), then we have the complete three-step process of natural selection (Dennett 1995). From here novel designs and truly new information emerge. None of this can happen without copying.
Sue Blackmore at The Stone
I want to make three arguments here.

The first is that humans are unique because they are so good at imitation. When our ancestors began to imitate they let loose a new evolutionary process based not on genes but on a second replicator, memes. Genes and memes then coevolved, transforming us into better and better meme machines.

The second is that one kind of copying can piggy-back on another: that is, one replicator (the information that is copied) can build on the products (vehicles or interactors) of another. This multilayered evolution has produced the amazing complexity of design we see all around us.

The third is that now, in the early twenty-first century, we are seeing the emergence of a third replicator. I call these temes (short for technological memes, though have considered other names). They are digital information stored, copied, varied and selected by machines. We humans like to think we are the designers, creators and controllers of this newly emerging world but really we are stepping stones from one replicator to the next.

As I try to explain this I shall make some assertions and assumptions that some readers may find outrageous, but I am deliberately putting my case in its strongest form so that we can debate the issues people find most interesting or most troublesome.

Some may entirely reject the notion of replicators, and will therefore dismiss the whole enterprise. Others will accept that genes are replicators but reject the idea of memes. For example, Jablonka and Lamb ( 2005) refer to “the dreaded memes” while Richerson and Boyd (2005), who have contributed so much to the study of cultural evolution, assert that “cultural variants are not replicators”. They use the phrase “selfish memes” but still firmly reject memetics (Blackmore 2006). Similarly, in a previous “On The Human” post, Benzon explains why he does not like the term “meme”, yet he needs some term to refer to the things that evolve and so he still uses it. As Wilkins points out in response, there are several more classic objections: memes are not discrete (I would say some are not discrete), they do not form lineages (some do), memetic evolution appears to be Lamarckian (but only appears so), memes are not replicated but re-created or reproduced, or are not copied with sufficient fidelity (see discussions in Aunger 2000, Sterelny 2006, Wimsatt 2010). I have tackled all these, and more, elsewhere and concluded that the notion is still valid (Blackmore 1999, 2010a).

So I will press on, using the concept of memes as originally defined by Dawkins who invented the term; that is, memes are “that which is imitated” or whatever it is that is copied when people imitate each other. Memes include songs, stories, habits, skills, technologies, scientific theories, bogus medical treatments, financial systems, organisations – everything that makes up human culture. I can now, briefly, tell the story of how I think we arrived where we are today.

First there were genes. Perhaps we should not call genes the first replicator because there may have been precursors worthy of that name and possibly RNA-like replicators before the evolution of DNA (Maynard Smith and Szathmary 1995). However, Dawkins (1976), who coined the term “replicator”, refers to genes this way and I shall do the same.

We should note here an important distinction for living things based on DNA, that the genes are the replicators while the animals and plants themselves are vehicles, interactors, or phenotypes: ephemeral creatures constructed with the aid of genetic information coded in tiny strands of DNA packaged safely inside them. Whether single-celled bacteria, great oak trees, or dogs and cats, in the gene-centred view of evolution they are all gene machines or Dawkins’s “lumbering robots”. The important point here is that the genetic information is faithfully copied down the generations, while the vehicles or interactors live and die without actually being copied. This interesting distinction becomes important when we move on to higher replicators.

So what happened next? Earth might have remained a one-replicator planet but it did not. One of these gene machines, a social and bipedal ape, began to imitate. We do not know why, although shifting climate may have favoured stealing skills from others rather than learning them anew (Richerson and Boyd 2005). Whatever the reason, our ancestors began to copy sounds, skills and habits from one to another. They passed on lighting fires, making stone tools, wearing clothes, decorating their bodies and all sorts of skills to do with living together as hunters and gatherers. The critical point here is, of course, that they copied these sounds, skills and habits, and this, I suggest, is what makes humans unique. No other species (as far as we know) can do this. Song birds can copy some sounds, some of the other great apes can imitate some actions, and most notably whales and dolphins can imitate, but none is capable of the widespread, generalised imitation that comes so easily to us. Imitation is not just some new minor ability. It changes everything. It enables a new kind of evolution.

This is why I have called humans “Earth’s Pandoran species”. They let loose this second replicator and began the process of memetic evolution in which memes competed to be selected by humans to be copied again. The successful memes then influenced human genes by gene-meme coevolution (Blackmore 1999, 2001). Note that I see this process as somewhat different from gene-culture coevolution, partly because most theorists treat culture as an adaptation (e.g. Richerson and Boyd 2005), and agree with Wilson that genes “keep culture on a leash.” (Lumsden and Wilson 1981 p 13).

Benzon, in responding to Railton’s post on The Stone, points out the limits of this metaphor and proposes the “chess board and game” instead. I prefer a simple host-parasite analogy. Once our ancestors could imitate they created lots of memes that competed to use their brains for their own propagation. This drove these hominids to become better meme machines and to carry the (potentially huge and even dangerous) burden of larger brain size and energy use, eventually becoming symbiotic. Neither memes nor genes are a dog or a dog-owner. Neither is on a leash. They are both vast competing sets of information, all selfishly getting copied whenever and however they can.

To help understand the next step we can think of this process as follows: one replicator (genes) built vehicles (plants and animals) for its own propagation. One of these then discovered a new way of copying and diverted much of its resources to doing this instead, creating a new replicator (memes) which then led to new replicating machinery (big-brained humans). Now we can ask whether the same thing could happen again and — aha — we can see that it can, and is.

A sticking point concerns the equivalent of the meme-phenotype or vehicle. This has plagued memetics ever since its beginning: some arguing that memes must be inside human heads while words, technologies and all the rest are their phenotypes, or “phemotypes”; others arguing the opposite. I disagree with both (Blackmore 1999, 2001). By definition, whatever is copied is the meme and I suggest that, until very recently, there was no meme-phemotype distinction because memes were so new and so poorly replicated that they had not yet constructed stable vehicles. Now they have.

Think about songs, recipes, ways of building houses or clothes fashions. These can be copied and stored by voice, by gesture, in brains, or on paper with no clear replicator/vehicle distinction. But now consider a car factory or a printing press. Thousands of near-identical copies of cars, books, or newspapers are churned out. Those actual cars or books are not copied again but they compete for our attention and if they prove popular then more copies are made from the same template. This is much more like a replicator-vehicle system.

Of course cars and books are passive lumps of metal, paper and ink. They cannot copy, let alone vary and select information themselves. So could any of our modern meme products take the step our hominid ancestors did long ago and begin a new kind of copying? Yes. They could and they are. Our computers, all linked up through the Internet, are beginning to carry out all three of the critical processes required for a new evolutionary process to take off.

Computers handle vast quantities of information with extraordinarily high-fidelity copying and storage. Most variation and selection is still done by human beings, with their biologically evolved desires for stimulation, amusement, communication, sex and food. But this is changing. Already there are examples of computer programs recombining old texts to create new essays or poems, translating texts to create new versions, and selecting between vast quantities of text, images and data. Above all there are search engines. Each request to Google, AltaVista or Yahoo! elicits a new set of pages — a new combination of items selected by that search engine according to its own clever algorithms and depending on myriad previous searches and link structures.

This is a radically new kind of copying, varying and selecting, and means that a new evolutionary process is starting up. This copying is quite different from the way cells copy strands of DNA or humans copy memes. The information itself is also different, consisting of highly stable digital information stored and processed by machines rather than living cells. This, I submit, signals the emergence of temes and teme machines, the third replicator.

What should we expect of this dramatic step? It might make as much difference as the advent of human imitation did. Just as human meme machines spread over the planet, using up its resources and altering its ecosystems to suit their own needs, so the new teme machines will do the same, only faster. Indeed we might see our current ecological troubles not as primarily our fault, but as the inevitable consequence of earth’s transition to being a three-replicator planet. We willingly provide ever more energy to power the Internet, and there is enormous scope for teme machines to grow, evolve and create ever more extraordinary digital worlds, some aided by humans and others independent of them. We are still needed, not least to run the power stations, but as the temes proliferate, using ever more energy and resources, our own role becomes ever less significant, even though we set the whole new evolutionary process in motion in the first place.

Whether you consider this a tragedy for the planet or a marvellous, beautiful story of creation, is up to you.


Aunger, R.A. (Ed) (2000) Darwinizing Culture: The Status of Memetics as a Science, Oxford University Press

Benzon, W.L. (2010) Cultural Evolution: A Vehicle for Cooperative Interaction Between the Sciences and the Humanities. Post for On the Human

Blackmore, S. 1999 The Meme Machine, Oxford and New York, Oxford University Press

Blackmore,S. 2001 Evolution and memes: The human brain as a selective imitation device. Cybernetics and Systems, 32, 225-255

Blackmore, S. (2006) Memetics by another name? Review of Not by Genes Alone by P.J. Richerson and R. Boyd. Bioscience, 56, 74-5

Blackmore, S. (2010a) Memetics does provide a useful way of understanding cultural evolution. In Contemporary Debates in Philosophy of Biology, Ed. Francisco Ayala and Robert Arp, Chichester, Wiley-Blackwell, 255-72

Blackmore (2010b) Dangerous Memes; or what the Pandorans let loose. In Cosmos and Culture: Cultural Evolution in a Cosmic Context, Ed. Steven Dick and Mark Lupisella, NASA 297-318

Dawkins, R. (1976) The Selfish Gene Oxford, Oxford University Press (new edition with additional material, 1989)

Dennett, D. (1995) Darwin’s Dangerous Idea, London, Penguin

Jablonka, E. and Lamb, M.J. (2005) Evolution in Four Dimensions: Genetic, Epigenetic, Behavioral and Symbolic Variation in the History of Life. Bradford Books

Lumsden, C.J. and Wilson, E.O. (1981) Genes, Mind and Culture. Cambridge, Mass., Harvard University Press

Maynard-Smith, J. and Szathmáry, E (1995) The Major Transitions in Evolution. Oxford, Freeman

Richerson, P.J. and Boyd, R. (2005) Not by Genes Alone: How culture transformed human evolution, Chicago, University of Chicago Press

Sterelny, K. (2006). Memes Revisited. British Journal for the Philosophy of Science 57 (1)

Wimsatt, W. (2010) Memetics does not provide a useful way of understanding cultural evolution: A developmental perspective. In Contemporary Debates in Philosophy of Biology, Ed. Francisco Ayala and Robert Arp, Chichester, Wiley-Blackwell, 255-72

37 comments to Temes: An Emerging Third Replicator

  • William Mullins

    Dr. Blackmore is clearly in thrall of imitation as the sincerest form of flattery and this attempt to extend the concept of replication into a third order of complexity in process preserving information is an understandable projection of the memetic construct. It is well worth chewing upon.

    If I’m understanding the origins of this proposal it clearly represents commitment to the dynamic of fully reductionist, molecular-based determinism. With that foundation the proposal begs a question or two: What other proposals in this “domain” are in competition with this one (e.g. Ulanowicz’ Third Window or Kauffmann’s Fourth Law of Thermodynamics? In which order will natural selection be anticipated to make the determination and by what agency?

    With respect to such competition, I can’t help but wonder how ready the presumed higher order species of “internet” is to recognize the issue let alone referee it. That presumes of course that some form of self-awareness is integral to the third order as described in the same fashion it is in the second order (i.e.humanity). Perhaps that is not the case.

    Before shrinking into subservience of constructed artifacts I would like to know that we humans have exhausted the possibility of our cognitive insight to comprehend how speciation and its “internet” counterpart comes about – something I believe the Dawkins School has yet to accomplish.

    • Thanks for all the comments. I want to respond briefly to some of them now, before tackling the major criticisms and then writing an overall response next week.
      William Mullins suggests I am in thrall to imitation.
      I guess so! Ever since I reread the Selfish Gene in 1995 and suddenly “got” it, I saw that evolutionary processes depend on having a copying mechanism and here is one.
      My take is certainly reductionist. I find bottom-up explanations fascinating and deeply satisfying but I am no slave to reductionism. It has its place and I think memetics is a good place for it.
      Where I completely disagree is over self-awareness which I think is irrelevant both to our evolution and to that of any higher replicator. What matters is what we copy or don’t copy – not whether we consider ourselves conscious or self-conscious when doing it.

  • Robert Faught

    memes and temes exist in a kind of symbolic information world in memory separate from human consciousness. A kind of collective unconscious. They have no more power over the human spirit than superstition. They are what naturalists use to try to create the human soul when they don’t believe in the soul. In the materialist view there is no real human person. We are just machines. Of course they imagine a type of imaginary collective unconscious ghost in the machine (society or the collective of digital machines) which is their definition of human beings in the first place. What if there is a soul or a person that is an eternal entity of some sort? then all your speculation and fear and anxiety about being controlled by a machine running amok is just childish superstition. When you take the naturalist primary assumption to its conclusion, the universe and human beings are an absurd childish nightmare.

    • “What if there is a soul or a person that is an eternal entity of some sort?” If so then I am wrong in almost everything I have ever written.
      But you are wrong, in my opinion, about memes and temes. They are nothing like a collective unconscious. They are real words, stories, skills, habits, ways of doing things, languages, religions, financial systems and many more aspects of our lives that have great power over us.
      As for my supposedly childish “fear and anxiety”, I am not talking about machines running amok and harming us, but about vast and powerful distributed processes. Scary they may be, but surely we are better to try and understand them.

  • David Ellis

    Having finished Jablonka and Lamb (2005), and currently struggling through Richarson and Boyd (2005), this article brightened my otherwise dreary Vermont morning. Thanks to the the author, the “Grey Lady”, and the internet, I was able to connect a few more dots. Articles like this crystallize vague, nascent feelings. Names emerge — Temes, Third Replicator. Once given a name, a complex idea becomes easier to discuss.

    Internet capabilities like Search and Wikipedia make it possible to dig down, and dig, and dig, and connect dots in ways I couldn’t imagine 15 years ago. Third Replicator indeed!!

    • I’m glad this brightened your day – though those are both very good books.
      But does your digging yet entail temes? After all you, the human being, are still the one wanting and choosing the information. The crucial step will be when artificial systems are out there doing similar digging for their own purposes. Actually they may already be but we don’t notice them (yet).

  • Jim

    You mention a few times that evolution copies the instructions for the product, not the product itself. An individual of a species is merely a vehicle for the information by means of which it is produced. This does not change fundamentally when we jump from genes to memes to temes.

    Yet it is curious that one of the most pervasive and successful memes – and one of the most remarkable accomplishments of evolution generally – is the idea of the individual person and the rights and respect we attribute to him/her. Subjectivity seems to be more than a statistical outlier to cultural evolution. In some respects our legal system depends upon it, since the suffering of the victim plays an important role in determining what is right and what is wrong, what is just and what is unjust.

    I wonder what the future of personality, subjectivity, and individuality are. As you point out, and as others before you have pointed out, evolution is not about the individual but about the species; or it’s at least about the instructions for producing the individuals, not those individuals themselves. Yet as we merge with our technology, we’re going to gain direct control over the evolutionary process. Gerhard Lenski hypothesizes such an outcome in his book on macrosociology. We’re not just setting temes free to do their own thing. To some degree, technology reflects our values and our emphasis on individuality and personhood. I wonder if the end result of this stage of the evolutionary process will be to undo the insignificance of the individual and to put the individual first as the ultimate, conscious goal of development.

    • A small correction to your first comment – I mention that biological evolution mostly uses copy-the-instructions. I suggest (in The Meme Machine and elsewhere) that memes are way behind in this respect and only rarely use this (better) method. Temes are even newer and further behind in this respect.
      You raise interesting questions here about the individual person and its current relevance to our legal system. I think we have our notions of responsibility and the person completely upside down – but that’s another story. (see http://www.susanblackmore.co.uk/research.htm#Free_Will)
      I do not see any reason why “as we merge with our technology, we’re going to gain direct control over the evolutionary process.” You cannot gain control over a vast dynamic process from the inside. Certainly we will affect it, as we are now doing with biological evolution through genetic engineering, cloning and so on, but we can never control it. I guess this is part of the point of all I have written – that the driving force for design and creativity is the mindless evolutionary algorithm. We need to be realistic about our role. Our consciousness (whatever that turns out to be) is irrelevant. For the first replicator we are a vehicle or interactor, for the second we are the copying and selecting machinery, for the third we are a stepping stone to the next phase. We may end up as little more than the energy providers, sucked into the larger system much as mitochondria were once taken into other cells to act as energy suppliers. We’ll see ….

  • The typical thinking on memes (this article included) misses the point.

    The main point of memes is that – by pulling us into imitation-type behavior – they prevent us from expressing our own individuality. While memes may be furthering their own evolution, they actually hinder ours. Dr. Blackmore’s theory won’t even consider this. In her model, the individual is denied status as an evolving entity (it’s only a “vehicle”). So if there’s no individual, then what’s left to evolve? Only memes and genes, of course.

    That said, the idea of memes is still valid. It’s just that in the usual bottom-up approach to defining them – when everything from songs, jokes and toe-tickling is a meme – we lose the big picture. In the work I’m developing at http://www.auria.org, memes are looked at from the top-down. In other words, memes are more akin to “cultural forces,” and songs and jokes are just the reflections (meme bits) of these cultural forces. What’s important about the top-down approach is that it highlights the “pull of the group” effect in memes (when we “imitate” others in the group), which runs counter to our evolution as individuals.

    In this vein, I’ve come up with a (working) re-definition of meme: an entity which influences or aligns our actions, outlook or behavior with that of a certain group.

    We’ve accepted the reductionist theory of memes too easily. It’s time to rethink.

    For more on this theme, see:

    What is a Meme?
    Why Memes are Important

    • You won’t be surprised to learn that you and I disagree completely.
      Memes do not prevent us from expressing our own individuality – rather, the reverse. We are far more diverse as individuals because each of us adopts, chooses and spreads different memes. Compare us with chimpanzees, crows, dolphins or any other intelligent species. Each individual is different by virtue of their different genetic makeup and what they have learned, but the differences are far smaller than ours because they have few, if any, memes. They do not speak different languages, wear different clothes, like different music or art, or take up different occupations and hobbies.
      You say that I deny the individual status or even existence but this is a complete misunderstanding. Of course there is an individual – a unique body with a unique collection of memes and a massive selfplex. We evolve, as do all plants and animals, but we have both genes and memes as replicators underlying the process.
      It is always worth reconsidering definitions but in this case I think you have completely lost the whole point of memes – that they are replicators. What you are talking about is, and has long been, part of social psychology, not memetics.

  • Nullius

    A very thought-provoking article. The question for me is what is the environment that selects between different temes, and how does it change? Where, in other words, are the selection pressures for these waves of data to be found?

    I propose we use the term “Cognitive Selection” to describe the culling operator in temetics. With genes it is the physical environment that sets the bar for existence; in sexual selection it is a set of genes in the competing animals or potential mates; and with memes it is the wetware of human neurology, and in particular mirror neurons.

    Today most “cognitive” processes are still considered to occur in brains, even though machines now hugely out-perform any biological brain in many domains. Before long, however, machines may well exceed our abilities in most respects by a massive margin. Already SETI researchers are starting to look for signals that originate from “machine intelligences”; as SB says, pretty soon most “cognition” will be performed by computers.

  • Susan Blackmore starts a good discussion on the notion of a teme. I have two items of feedback:

    1. Is imitation really the full key?
    I think imitation is one part of the key, but social culture only took off because of language and ultimately the ability for abstraction.
    Abstraction makes all the difference: the ability the refer to things that are not literally things. One big other step was the evolution of written language. This written language now also plays a huge part in the proposed temes: many layers of abstraction stand in between humans and what computers do (or let robots do), these layers are ultimately all built up by abstract language-structures.

    2. Are temes special in a way?
    I am not at all shocked by the proposal of using evolutionary theory to explain human culture. including modern forms of  culture like in economic (automated) processes. I disagree with William Mullins that this is either deterministic or reductionist: there is no reason why a world built of organisms cannot be dynamic of holistic.

    I am triggered by the idea of constraints: human individuals are very much constraint by their poor ability to understand parts of what is out there information wise. Non of us can know and understand all the theories, models and data out there. Somehow culture and information and technology is bigger than any of of us. We are also born in a world with many institutions, laws and other cultural memes out there that create their own world an individual cannot escape from.

    The memes making up our constitutions and governmental bodies, much like religions, make sure that people under their “jurisdiction” will support them. There are good reasons for it in the sense of they are often “for the common good”. But such institutions also create a wild west of memetic competition. Because humans do not oversee all of it many things in social processes and social institutions end up going wrong. Well, not really going wrong, because there was no objective “right” to begin with of course.

    Susan Blackmore refers to technology making (copying) things that humans have little influence over. I see the same happening in organizations and cultural institutions and systems: many things happen that are unforeseen and unwanted.
    So in the end I am not sure if there is anything really special about temes I have not grasped. Why are they different form institutional memes? They exist because they are useful to humans, and yes can get out of control or be a part of processes that get out of control and thus become harmful. But so can institutions under the influence of meme-human gangs like fundamentalists in politics, religion or any other socials stage.

    I do not object to the notion of a teme, because obviously technology is memetic and vastly important to understanding our world today: socially and economically. But I see no fundamental “third” layer here (yet)?

    Hans-Cees Speel

    • 1. You are right, imitation is far from the “full key”. The reason I emphasise it is to stress that a new kind of copying is essential for an evolutionary process to get off the ground. More is needed, but copying is essential. Imitation is the new kind of copying that let loose memes. Abstraction and language (in my opinion) followed naturally from that beginning.

      2. Here you make many good points, but I am sorry that I have not made my own argument clear enough. I will try to do so now.
      I too have wondered (going round and round for many years and sometimes getting very frustrated) whether all this digital information should be considered as just more memes (i.e. the same replicator taking new and varied forms) or whether something fundamentally new is going on.
      My intuition suggested something new but logic found no reason why. Eventually I decided to go back to first principles (indeed, as Dawkins advocated when he coined the term meme). The crucial principle is that you have new evolutionary process when you have not only a new kind of information but a new means of copying that information with variation and selection. i.e. all three of the critical processes must be in place.
      Now it seems obvious that the “institutional memes” you mention are indeed still memes (whether harmful, helpful, under our control or not). We do (or will) have a new replicator only when computers, handling digital information, are also varying and selecting that information. Is this happening? I suggest we see the beginnings of this in search engines, some anti-spam processes, free-roaming bots, some worms and viruses that are abandoned by their creators, and maybe more. We might not even know how much of the vast reserves of storage space are occupied by information that has never been seen by a human, never will be, and yet is available for selection by machines.
      Once this principle seemed clear to me I wondered whether the step from memes to temes would be anything like the previous step from genes to memes. I could now see that it looks the same. In the first case a product (vehicle or interactor) of the first replicator became copying machinery for the second one. Now computers constructed as useful products for us (vehicles carrying, storing and copying memes) became capable of carrying out variation and selection on their own. The crucial step is indeed of the same kind. The vehicle of one replicator becomes the replication machinery for the next. The way this became obvious to me suggested that going back to first principles was the right way to go and that this justified me in giving a new name to the new process and the new replicator.
      I do hope this makes clearer what I have been trying to do. I would welcome your further critique and thoughts on this.

  • Dave McGown

    My lazy brain is wont to accept “temes” as readily as my ancestors accepted spirits and gods. (“I’m sorry, Dave, I can’t do that.” -HAL) While scientists strain to create “new” elements that exist for billionths of a second, species of life wink in and out of existence. So far, humanity is a barely-noticeable bump in the road for cockroaches. I can think of no reason to doubt the inevitable ascendence of this Third Replicator, nor a Fourth…et cetera. After all, an Infinite Universe will accomodate. Yet, we parce our language and struggle to be “definitive” in a self-referential attempt to control, or at least to understand that which is merely a reflection of our own memes, or thinking. Still, it is entertaining to imagine the time when our species (flesh & blood)has expired, leaving behind something else (temes?) that carry on. I wonder what “they” will come up with?

  • Whilst attracted by the elegance of a third replicator I believe the word ‘teme’ needs a more rigorous definition in order to justify its coinage. Toward that end it’s instructive I think to look more carefully at the definition of ‘meme’. Dawkins defined it as “a unit of cultural transmission” and – in the context of this response – the important word is ‘cultural’. Memes then, are the informational ‘quanta’ of cultural transmission, and the value of the word meme is that it helps us to think about and debate cultural evolution. Does it make sense to ask what temes might be the informational quanta of? I believe so.

    What then, is the ‘temetic’ equivalent of culture? Dawkins talks of “the soup of human culture” and it’s easy to think of a new kind of ‘digital’ soup – all the stuff in the Internet – for example. But I’m not convinced that this digital soup is sufficiently differentiated from human culture to justify ‘teme’. After all most of the stuff in the Internet is just digital representations of good old fashioned human cultural things, like newspapers, memos, gossip, books, music, photos, and videos. A cool idea propagated by Facebook, is – I would argue – still a Dawkinsian meme. The reason such memes can spread fast is surely as much to do with the pervasive communications network of the Internet as it is to do with high fidelity copies.

    Susan refers to the search engines – without doubt the most powerful tools on the Internet. Think of the Web crawlers that tirelessly trawl web pages in order to build the indices that allow us to search for stuff with breathtaking speed. Perhaps the meta-knowledge in those indices – generated autonomously by computers for the exclusive use of computers – is the temetic soup we’re looking for? But the problem here is that – despite appearances – the Internet is highly heterogeneous. The machine-generated meta-knowledge in Wikipedia is certain to be of a completely different form to that of Facebook, or of Google. Does it even make sense to think of many kinds of temetic soup, each alien to the other, and all alien to humans?

    The reason that I think a rigorous and useful definition for ‘teme’ eludes us, and will continue to do so for some time is this. Culture, as a phenomenon, existed long before Dawkins came up with the idea of the meme. The problem is that we don’t yet have a technological phenomenon of sufficient power, autonomy and singularity that it demands a descriptor equivalent to culture. Until we do, arguing about temes may be too much like trying to put the epistemological cart before the horse.

    Postscript: In the Bristol Robotics Lab, UK, there is a group of mobile robots that are arguably the closest yet to ‘teme’ machines. They are autonomous* robots that imitate each other’s behaviours and, because the robots can only see and hear each other with their physical sensors, imitation is imperfect. Thus we see variation in imitated behaviours, selection (because robots choose which observed behaviours to enact) and inheritance (because we see inherited characteristics in n-th generation imitated behaviours). In short, we observe memetic evolution. The robots were designed and built as embodied models of meme machines, but we could also think of them as not models of anything but teme machines in their own right. *and in case anyone is worried, their batteries only last about an hour and a half.

    • Alan Winfield raises the important question of definitions. I don’t believe I have ever offered a formal definition of “teme” and perhaps it’s time that I did.
      Here’s my first attempt. Teme: information copied, varied and selected by machine.
      Here’s another attempt. Teme: the replicator involved when machines copy, vary and select digital information.
      I might need to point out that I am excluding human beings from counting as machines in these definitions, or I might need to specify the kind of machine (computer? interlinked set of computers?) but I don’t want to limit it too much. I think the essential point is clear in both. A machine that was originally created by and for humans has begun carrying out all of the three processes necessary for a new level of evolution with a new kind of replicator. The machine has become a teme machine and the replicator the teme.
      “What then, is the ‘temetic’ equivalent of culture?” asks Winfield. Not, I suggest, “all the stuff in the Internet”. Given my reasons for coining the term in the first place this should be clear. Most of the stuff in the Internet (at the moment nearly all of it) was put there by and for people. People wrote the texts, took the photos, wrote the songs and then other people chose, or not, to have them copied onto their computers and so caused them to spread. The answer to any question of the form “Why was this meme successful?” is nearly always “Because someone liked it, found it useful or funny or in some way interesting enough for them to tell other people about it”. So I agree with Winfield that all this is “Dawkinsian memes”.
      This is what I think is going to change as machines begin adding variation to the information they copy, and selecting it without our involvement. Right now I guess the amount of stuff out there that is genuinely copied, varied and selected by machines is tiny but growing. So my answer is that the “temetic soup” is this information. This is what we need to watch. But how? Do we know, even now at the beginning, how much storage space out there is taken up by this kind of information? How much processing capacity is being used this way? Could we find out? I would welcome any expert who could tell me more about this.
      Postscript: I have watched the development of Winfield’s imitating robots with delight, and have also thought of them as meme machines, but I am tempted by his new assessment. Defining temes as in 1. above these little robots may indeed be among the first autonomous teme machines; defining them as in 2. above they would not count because their memes (or temes) are not digital information but actions and sounds made by the little robots. What do you think Alan? Keep the word “digital” in the definition or leave it open to other systems and your robots too?

  • Nosyrev Ilya

    A very interesting article, one more important step on the way of understanding what replicators are. I think that the computer programs of the future, much more complex than modern, will indeed be very much alike to memes – because the processors and operational systems will come closer to the abilities of a human mind.

    The question that comes to me is: maybe, cars and books are not worse than computer programs and also can be named temes? Let me give some reasons.

    Actually, all objects created by humans, starting from the Stone Age, demonstrate the principles of Universal Darwinism. For example, stone axes that archaeologists find on a primitive men’s site, are very alike and at the same time show gradual elaboration of more optimal forms. The principles of heredity, changeability and natural selection here are so clear that if the archaeologists were extraterrestrials they could take these objects for fossils of some animals. But can we say that copying these objects is possible due to copying memes that bear the technology of creating a stone axe? No, indeed people actually copied only 2-3 basic methods of chopping off fragments, that let them create all the diversity of stone tools. And copying of an axe was being done that way – a man gave a look at a model, then split a fragment from his stone, then again gave a look to the model, then one more split – and so to the end of the job. It’s the same way that a painter copies a picture – he doesn’t copy memes of the artist that created the original – he just analyzes the original and makes his own variant of the picture with the help of his own memes, born from this analyses. Human beings are the machines that can reproduce different kinds of objects, but this copying doesn’t mean a copying of memes – different people reproduce the same object using different memes and human ability to imitate. Transfer of memes from a man to a man is necessary only if the technology of fabricating a product is rather complex, but even in this case memes serve as mediators between human brain and the products that human make, and some of these products can piggy-back on both memes and people (as religious symbols – Christian crosses, Buddhist stupas, etc., that change religious mythology in the way that makes people reproduce still new copies of them). But in most cases in the history of technology we can see collaboration or even coevolution among people, memes and artificial objects, which are a rather selfish player in this game.

    What are being replicated in the examples above? Stone axes? Pictures? Are they replicators? Probably, no. What is replicated, is a FORM, the system of mutual relations between elements of a unit. I think, that form is the widest notion for all the things that can replicate. Genes, memes, computer programs, pictures, cars, produced by factories, are forms, because they are the products of copying not substance, but information. Probably, many different kinds of objects in the Universe potentially can become replicators – if there are necessary conditions. Your example of computer programs as replicators is brilliant, partly because it shows a class of objects that can use various vehicles: hard drives, flash-memory, laser disks and paper punch cards are physically different, but a program can be exactly transfered to each of this stores, because it preserves the integrity of its data. This example demonstrates that material of which a replicator is made doesn’t matter much – so cars and books made of solid materials, genes consisting of nucleotides, computer programs stored on silicon plates and memes stored in our minds, are things of the same rank: only thing that replicates is their form. Every class of these things has its peculiarities: programs are replicated exactly, but it’s not clear if they can mutate with no help of a programmer; objects produced by humans are copied approximately and in fact have no vehicle; questions about memes’ replicating are not yet answered, etc.

    In your opinion, is such an approach have a right to exist? One reason for me to like it is that it can give the final materialist answer to the philosophical problem of universals, raised by Plato and still actual in modern interpretations.

    Sincerely yours,
    Ilya Nosyrev

    • Ilya Nosyrev asks about stone tools, cars and books. Here we are in the realms of that long argument concerning whether memes are inside human heads or outside of them and whether there are meme products (phemotypes) as well as memes. I do not wish to delve into these disputes again here (but see Benzon below). I will just point out that whichever view you take Nosyrev is right that stone axes, cars and books all have similar status, as do computer programs. The big difference – the step to temes – comes when one of those products becomes capable of copying with variation and selection. This has not happened with most computer programs but is, I suggest, beginning to happen now.

  • Neil Pundit

    Susan Blackmore seems brilliant but utterly confused and carried-away. There is nothing wrong in her observations. It is when she starts theorizing I am baffled to say the least. Imitation, copying, replicator, .. have been part of our nature (culture and civilization). What is important is that we can add original ideas to improve or for a new design. (Mere copying does not increase the information content.)

    Mass production of goods is “copying” at a reduced cost. Copying of information is progressively becoming inexpensive. Internet is helping with a wider dissemination, just as books did. Guidance to digest, interpret, and apply – is a different matter. Gurus, teachers, philosophers do that.

    Copying, adaptation, evolution, .. have been at work in biology, botany, and zoology. Gene research may enable us to better fight diseases, overcome our physical limitation or ailments. Well and good.

    So, what is Susan worried about? End of the world by uncontrollable copying on the internet, misguided gene mutation, or whatever!! But certainly we don’t need a new term from her or her associates to understand what we already understand well enough.

    - Neil Pundit

    • I agree that “Mere copying does not increase the information content”: copying with variation and selection does. “Original ideas” are recombinations or variations of old ones. This is the whole point.

  • Three things:

    1. First, a comment I made in the discussion at the New York Times, and which Blackmore acknowledged: She has misstated my problem with memes, asserting that it is the TERM I do not like. Not so. I think the term is brilliant, which is why I use it. What I object to is the USE of the term to indicate quasi-autonomous bots that go hopping about from brain to brain commandeering neural realestate in competition with other one another.

    Beyond that, it’s clear that we think about memes, and cultural evolution, in rather different ways. My current views, with at least one fairly detailed example, Rhythm Changes, can be found in this set of notes. My position has changed a bit since I posted here a month and a half ago and I’ve written a post about that; that post is included in the current version of those notes.

    2. It’s interesting to think about these three realms, biology, culture, and digital technology, in terms of how well we understand them. Clearly we understand digital technology the best. After all, we created that realm, and with great deliberation. We have substantial understanding how those things work, though there are certainly problems. The annoying persistence of bugs in software indicates a fairly substantial gap in our understand of how this technology works and that despite the fact that we’ve created it.

    Next comes biology. Culture is last. While there is substantial disagreement among biologists about the details of evolution (e.g. what, exactly, is a gene? a species?), students of human behavior do not share any single explanatory scheme comparable to that of evolutionary theory in biology. In particular, the human sciences have not adopted an evolutionary framework for human culture.

    This would suggest that, if temes are indeed a phenomenon of digital culture, that our understanding of them should come quite rapidly as our understanding of the underlying substrate, though not at all complete, is more substantial than our understanding of the biological and cultural realms.

    3. Third, temes would seem to require substantial autnomous operation among teme machines. It’s not at all clear to me that we have that. But if you want a concrete fantasy in which to construct some thought experiments, here’s a blue sky proposal that NASA published in 1980. It’s a “proposal for a self-replicating automated lunar factory system, capable of exponentially increasing productive capacity and, in the long run, exploration of the entire galaxy within a reasonable timeframe.” Surely such a thing would be teeming with temes.

    • 1. I am glad you like the term and am sorry about the misunderstanding.
      However, I do not think of memes as “quasi-autonomous bots” hopping from brain to brain but rather more simplistically as all the information we come across, day after day and have to deal with. Some we use and forget, some we effectively copy on to other people, most we do neither. This includes stories in the morning paper, adverts, emails, songs, jokes, theories, designs for buildings or tea pots, images, dance steps, and countless other things. “Commandeering neural real estate” makes it sound as though the neurons ought to be doing something else, but no, they are doing what they are designed to do, process all this stuff. Inevitably there is competition because there is just too much stuff for any single brain (piece of “neural real estate”) to cope with.
      More importantly – yes, you and I do think about memes rather differently. In the links you provide you discuss the much-debated question of whether memes are inside heads, outside heads, both or neither, and you explain why you have recently changed your mind. As you know this debate has gone round and round for thirty years. Dawkins too changed his mind from what is sometimes called “Dawkins A” to “Dawkins B” (memes in heads). My own view is that this whole muddle comes about because most memes (including your interesting example of Rhythm Changes) are not organised into any equivalent of a germ line/phenotype or replicator/vehicle distinction. I have explained this before (in The Meme Machine and elsewhere) in terms of the benefit of copy-the-instructions for making something compared with copy-the-product itself. I suggest that new replicators start out with the latter and evolve towards the former.
      Some memes (e.g. factory produced goods and some computer programs) do use copy-the-instruction systems. Only in these circumstances can we meaningfully ask whether the car, book, word processor or whatever is the meme or the meme product. For many other kinds of memes the question simply makes no sense. So this, in my opinion, is not a debate worth prolonging unless we can do research into how memes have made the transition to this better system and how and when it works.
      2. As I have explained in other responses, I think temes are only a tiny part of our current digital technology and this is part that we understand very poorly, and may always have trouble understanding since, by definition, it is information handled by machines and not necessarily seen by any human at all.
      3. Nice example, but here the human being are assumed to be still in overall control. I doubt this will now be possible.

      • Excuse me, but this is not a correct characterization of my view:

        My own view is that this whole muddle comes about because most memes (including your interesting example of Rhythm Changes) are not organised into any equivalent of a germ line/phenotype or replicator/vehicle distinction.

        In the case of music, a given performance (whether live or mediated) is the cultural phenotype or vehicle. Cultural selection acts on performances. If people like a performance they will want repetitions of it. Just what that means depends on the type of music and whether or not the particular performance was a recording or a live performance.

        A musical performance based on Rhythm Changes must necessarily involve many other memetic elements and many elements that are not memetic. There will be specific rhythmic and melodic elements, specific voicings, and so forth, all of which are necessary to have any performance at all. Rhythm Changes is not the whole performance; rather it is a fairly abstract and complex property that a performance can have.

        It would be physically impossible to have a performance consisting of nothing but Rhythm Changes. Such a performance would be roughly comparable to the Cheshire Cat’s grin without the cat. One can write about such things, one can even imagine them. But you’ll never see it in the real world.

      • I am sorry that you think I have incorrectly characterised your view. I do not believe that I have.
        You say that “In the case of music, a given performance (whether live or mediated) is the cultural phenotype or vehicle.”
        I disagree. I do not believe that in the case of music and performances it makes any sense to try to divide them into replicators and vehicles. You seem to think I imagine some weird kind of “physically impossible” performances, but this is unfair. Both my sister and nephew are jazz musicians so I have some idea what jazz performance means and how rhythm changes may appear. But which is the meme and which the vehicle?
        You say the performance is the vehicle but I claim that in the case of music, and many other kinds of performance and action, there is no clear distinction between replicator and vehicle. There is no equivalent of the germ line through which information is copied accurately, while producing vehicles which interact in the real world and determine copying frequencies. Music is simply not organised this way and we only get in a muddle (as the history of memetics shows all too well) if we try prematurely to divide human culture into replicators and vehicles along the lines of too close an analogy with biology.

  • Ron Randall


    It is not life that is evolving, it is knowledge.

    Life is just a temporary container for the knowledge (your “temes”) that will not die and have to be relearned or copied by others when it moves into the permanent, digital processor-memory of the evolving “web” or its successor incarnation. That knowledge can simply grow, with ever-increasing speed, without the need for replication.

    Raymond Kurzweil’s book “The Singularity is Near” clearly focuses on the inevitability of the human brain’s being surpassed as the high water mark of knowledge-carriage and even knowledge-creation. If a reader cannot imagine this happening in the 30 years he predicts, that reader cannot fail to see it as unavoidably likely in, say, the next 1000 years — a mere blip in the evolutionary timeframe.

    Since neither Kurzweil nor I can imagine what the world (or universe?) will be like after the “singularity” event, I would enjoy learning your thoughts on this…

    • Kurzweil’s view is somewhat different from mine in that he emphasises intelligence, and mostly talks about individual humans or combined human-machines. I am thinking more of evolutionary processes in cyber-space entailing replicators that build themselves entirely new kinds of bodies or virtual bodies, but there are otherwise many similarities between us.

  • Ronald Calitri

    In this discussion, meme and teme seem rather slippery concepts, probably because the concepts are inadequately specified. Let’s drill backwards. As an economist, I’m used to people who argue in terms of one “theory,” or another, but am hardly favorable to any theory unless the relations it describes are observable. The “data” for economics are steadily accumulating, not simply as time passes and more activities occur, but also as previously unused data recorded in the past are brought into the contemporary universe of available sources. Both new and novel data are studied using evolving methods. So, “we” economists are expressing a ‘meme,’ but conducting our real ‘business’ using ‘temes,’ exchanging, complimenting and competing, if I have understood these concepts correctly. Stepping backward through Professor Blackmore’s example, the automobile will be built by self-replicating robotic factories, is built by numerous assemblages of specialized workers and machines in a variety of production processes, and was built by mechanics of genius combining and assembling ‘temes’ from different predecessor ‘memes.’ But auto-makers are not ‘copying’ or ‘replicating,’ the physics, chemistry, biology principles embedded in their products. Their technique is a matter of adaptation, or exaptation not imitation. Similarly with economists, they engage in the recursion of previous observations against new data and thought experiments.

    But, where in all of this is the “meme” of the automobile? Are you going to accept my argument that economists share a meme? Yes, with a little conversation we could probably seize upon and communicate to each other a meme of either the automobile or the economist, drawn from two centuries of common experience with such entities. But, such memes would not express in any novel fashion the desirable concepts for these entities lying much deeper in historic consciousness. The automobile is simply a “chariot,” capable of swift flight through the common people. Anyone who has driven feels that ancient thrill; and no doubt the enjoined grandiosity is responsible for our reluctance to economically accommodate the polluting side-effects. So,in archaeological essence,the “auto” meme is about 3,500 years old. As to economics, it seems, primarily to have been invented by about 2,500 years ago in response to the effects on human happiness of the hierarchical social effects of such memes as chariots, land-ownership, credit, and elite profligacy that were widely observable by that time. At least, both Socrates (via Xenophon) and Aristotle left works titled “Economics.”

    I am in agreement with the primitive frame if not the noted method in the post above by Nosyrev Ilya. Clearly, stone tool manufacture was a social enterprise; and the skills of fabrication were learned by observation and experience, even discussion, though perhaps not in the modern-form language, we think to be not less than 50,000 years old. In terms of this discussion, we might wish to draw the line, sometime after then, but certainly earlier than 2,400 years ago, and the Athens “Academy,” and definitely no later.

  • Andrew Atkinson

    Having not read all the posts, I’ll venture to say that I don’t think it is fruitful to attempt to define temes in terms of memes or memes in terms of genes. The point is, that they are ‘kinds’ of replicators.

    The crux of the whole memetics argument, and which carries through in similar vein to the temes debate, is whether or not the notion of selfishness ultimately makes any sense – and it might not. However, what’s important about genes, is just as important for memes – that the information they hold ‘can’ indeed be transmitted faithfully down generations retaining both fitness increasing adaptations and peripheral viral noise. The two (genes and memes) do not behave the same way, but can, individually accumulate into more complex evolutionary entities.

    But temes? How, say, is a transistor radio, or iphone, transmitted from one generation to the next? – certainly not by any genetic means,and much more likley through culture. Aren’t then temes just the phenotypes of memes? Are temes just another product of evolution? his smack, Sue, of your soup analogy in the Meme Machine?

    The sinister notion of selfish replicating entities, beautifully illustrated by Dan Dennett’s tale of the Lancet fluke, is an interesting one – and certainly we are becoming ever more reliant on our ‘external brains’ (memory sticks, iphones, computers etc but, that this has become reason to suppose that we’re the sheep and the iphone the lancet fluke does take some ambitious reapplication of the concept of a selfish replicator.

    To think when we first saw the value of recording information in books to preserve knowledge and stop us forgetting things that the book was ultimately being selfish is a little unfair on the book.

    My feeling is that once we get very clear conceptually on the perennial problems in evolutionary theory itself and define replicators properly, and vehicles, and interactors, and individuals, and groups. and selfishness, and fitness, and phenotypes and genes, and another few hundred pages later, then technology, from spider web to world wide web, will likley fall into it’s proper place so considerd from an evolutionary perspective.

  • Robert Faught

    As I continue to listen to this debate, I hear no mention of the sentient organism or recognition of the human person (conscious subject) as having any significant role in evolutionary biology. There are genes, memes, and temes that have greater evolutionary significance than the organism or vehicle. These are biological, cultural, and informational replicators that are all defined as physical entities. Genes reside in physical bodies, memes reside in physical brains as memory, and temes reside in physical objects of technology such as computers, cars, etc. They are conceived as having an independent reductionist existence that transcends the individual organism, culture, or cultural/technological object. For this reason I would like to suggest a forth replicator. Before I do, I would like to set some foundation. For example, the computer has gone through an evolution in form from mechanical machines, to transistors and magnetic core memory, to integrated circuits. But the function of the computer remains the same; to process information. The computers that do the better job at processing information have greater survival value. The variations in design, materials, and physical construction are all secondary. However, no matter what the design of the computer, the computer cannot, of itself, understand anything. The information processed in the computer consists of symbols, language, algorithms, 1’s and 0’s. There is a relationship between this information and the behavior of machines and computers, but there is no understanding. In order for there to be understanding, there must be a ‘person’, an ‘I’, someone ‘who‘ understands. Understanding transcends language and symbols because understanding can only be held within and by a conscious subject. We all know that consciousness and mind cannot be measured by scientists yet, so nothing scientific can be said as to the real nature of the conscious subject. But we do know that understanding can only be held by a conscious subject. I would also say that the greater understanding an organism has of his environment (not just the greater quantity of information), the greater the survival value of the organism. Therefore, in addition to gene, meme, and teme as biological, cultural, and informational replicators – I think we should add a forth replicator which is the unit of understanding. Genius or insight is the accumulation of (units of) understanding upon understanding, not information upon information. I don’t know what this forth replicator should be called, but without it, I think the argument is ultimately meaningless.

    • Thanks for bringing up the role of the whole person or subject. People are, I would say, the vehicles of the genes, the copying machinery for memes, and the inventors of the original teme machines that may (or may not) now take off on this new evolutionary explosion.
      I disagree entirely about the relevance of either consciousness or understanding. Genius and insight are the results of a clever brain taking on lots of old memes and using them to construct new ones – whether by recombining them in new ways or altering them. I do not believe that consciousness plays any causal role in this (see http://www.susanblackmore.co.uk/Books/Consciousness/2Ed/index.htm for a summary of these arguments). As for understanding, much recent research suggests that machines are capable of understanding if sufficiently embodied or grounded in the physical world It is interacting with the world that matters, not something mysterious called consciousness, insight, or understanding.

  • Nosyrev Ilya

    I’d like to make my view clearer — all the tools that humans produce are capable of copying with variation and selection. I don’t mean serial production, I mean the step-by-step evolution of a tool, accommodating to both the task, set to it by a human, and to the physical reality. I think that artificial objects are not only physical embodiments of memes – they are a separate class of replicators, co-evolving in symbiosis with memes. But it’s another problem, you are right.

    As concerns computer programs, there is indeed a lot of examples of their selection. For instance, when you enter a query into a search engine (Google, etc.), it shows all found documents ranked according to relevance. The first items in this rank have much more chances to be load to the computers of millions of users. And some documents can “deceive” a search engine and reach the first places without being relevant – because search engines actually take for relevant the documents, that have a special structure, a successful disposition of key words in the text, etc. What is still more interesting, in this rank can be even a “trash text” – a fragment of code, accidentally generated by a program. The percentage of computer generated code in the Web even now is about 90% of all the data transferred through Internet, and some part of this “trash” really lives on the laws of Universal Darwinism. But such documents and computer viruses need a human-made operational system to be copied – they are rather passive (as biological viruses and artificial objects, produced by humanity – that was what I meant in my previous commentary).

    By the way, there is one more kind of temes, that can emerge in the near future. These are nanomolecular machines, able to reproduce themselves. They were described in Eric Drexler’s “Engines of Creation”, and, as some scientists predict, they may be created in 20-30 years. Such machines will have its own “genetic code”, according to which they’ll make new copies of themselves. An interesting fact – as physicists predict, from time to time the “genetic code” of some of these machines will be damaged because of solar radiation, magnetic fields, etc. For the most part, it will lead to a breakdown, but sometimes it may become a useful mutation that will give the mutated machine the advantages over others. It’ll be a real evolution of artificial objects.

    Problem of temes is very important, because the number of new replicators that will emerge due to technological progress in the near future, can be huge, and the humanity doesn’t realize how selfish they can be – even without being “living creatures” in the common sense of this term. I think that computer programs are only precursors of future powerful technological replicators.

  • I would like to thank everyone who has joined in this debate about the idea of a third replicator. Sadly, I fear my argument was not clear in at least one important respect. So I’ll try to put that right. Hans Cees Speel asks whether there is “anything really special about temes”, and sees no fundamental “third” layer. Ilya Nosyrev asks why stone tools, cars and books are not also temes. And Alan Winfield asks for a more rigorous definition of teme.

    As I explained in my reply to Speel, I have long been pondering whether all digital information is just more memes (i.e. the same replicator taking new and varied forms) or should be considered as something new. My intuition suggested something new but logic found no reason why. Eventually I tried going back to first principles (as Dawkins advocated when he coined the term meme). The key principle here is that evolutionary processes require variation, selection and heredity. So a new kind of evolution requires not just a new kind of information but a new means of copying that information with variation and selection. i.e. all three of the critical processes need to be in place.

    In the context of modern technology, this can happen only when computers are capable of carrying out variation and selection as well as storing and copying digital information. Is this happening? I suggest we see the beginnings of this in search engines, auto-generated spam websites, some anti-spam processes, free-roaming bots, some worms and viruses that are abandoned by their creators, and maybe more. We might not even know how much of the vast reserves of storage space are occupied by information that has never been seen by a human, never will be, and yet is available for selection by machines.

    This explains why I responded to Winfield’s challenge by coming up with the following possible definitions.

    Teme: information copied, varied and selected by machine.

    Teme: the replicator involved when machines copy, vary and select digital information.

    I might need to point out that I am excluding human beings from counting as machines in these definitions, or I might need to specify the kind of machine (computer? interlinked set of computers?) but I don’t want to limit it too much. I think the essential point is contained in both.

    Once this principle seemed clear to me I wondered whether the transition from memes to temes would resemble the earlier transition from genes to memes. It does. In the first case early humans, constructed as a product (vehicle or interactor) for genes, became capable of copying, varying and selecting a new kind of information – memes. Now computers constructed as products for us (vehicles carrying, storing and copying memes) have become capable of copying, varying and selecting a new kind of information. The crucial step is indeed of the same kind. The vehicle of one replicator becomes the replication machinery for the next. The way this became obvious to me suggested that going back to first principles was the right way to go and that this justified me in giving a new name to the new process and the new replicator. The machine is a teme machine and the replicator the teme. I do hope this makes clearer what I have been trying to do.

    I want to respond more briefly to some other points. William Benzon and Ilya Nosyrev both have problems with attempts to divide aspects of culture into replicators and vehicles, and the related question of whether memes should be considered to exist inside human heads, out in the world, or both, or neither. Benzon provides links to his work outlining why he has recently changed his mind.

    This is a debate that has gone round and round for thirty years. Dawkins too changed his mind from what is sometimes called “Dawkins A” (memes as skills, habits, technologies etc) to “Dawkins B” (memes as representations in heads). Others have changed their minds too, but going back to the origin of the term “meme” as meaning “that which his imitated” suggests that anything that is copied should count as a meme. Naturally this varies with the kind of behaviours, technologies or ideas that we are dealing with and there is no fixed answer that applies to all.

    My own view is that this whole muddle comes about because most memes (including Benzon’s examples from jazz music) are not organised into any equivalent of a germ line/phenotype or replicator/vehicle distinction. I have explained this in The Meme Machine and elsewhere in terms of the benefit of copy-the-instructions for making something compared with copy-the-product itself. I suggest that new replicators start out with the latter and evolve towards the former, as appears to have happened with the transition from RNA to DNA based life forms.

    Some relatively recent memes (e.g. factory produced goods and some computer programs) do use copy-the-instruction systems. Only in these circumstances can we meaningfully ask whether the car, book, word processor or whatever is the meme or the meme product (for an answer we can ask which is copied, which is visible for memetic selection). For many other kinds of memes the question simply makes no sense. So this is not a debate worth prolonging.

    In a similar vein Andrew Atkinson asks “Aren’t then temes just the phenotypes of memes?” The above should make it clear why I say “no”. You could say that the computers we have built are vehicles for our memes (in the sense that they carry, protect and store them) or even phenotypes of the instructions for building them, but temes (as I have defined them) exist only when those computers themselves begin to carry out all the three processes of variation, selection and heredity.

    Many people, both here and in responses to the same piece at The Stone (New York Times) posit a role for consciousness. William Mullins presumes “that some form of self-awareness is integral to the third order”, Robert Faught thinks that memes and temes form a “kind of collective unconscious”, and Jim argues for a “conscious goal of development”. I think consciousness is just a distraction here, not because I don’t think it interesting (I have devoted much of my working life to its tantalising mysteries) but because one of the joys of memetics is that you have to concentrate on who copies what to whom and why; it is entirely irrelevant whether they do this consciously or not. Teme machines or complexes of temes may, like us, become deluded into thinking they are a conscious self, but this would be a consequence of their own self-modelling rather than a driving force in the evolutionary process.

    I have much enjoyed being given this chance to try out a new idea and see what people think of it. The real test will come when we see whether the idea provides any useful testable predictions or helps us better understand what is going on. For now I hope we can continue debating whether the concept of a third replicator makes any sense or not.

    • Hans Cees Speel asks whether there is “anything really special about temes”, and sees no fundamental “third” layer. Ilya Nosyrev asks why stone tools, cars and books are not also temes.

      I’ve seen many people commenting on your articles (which I’ve been following for the past year) misunderstanding the “teme” concept in this same way.

      As someone with an academic background in both Linguistics and Computer Science and with an insatiable appetite for evolutionary science, it didn’t take me long to extrapolate from the meme into the teme, and understand that, somewhat like what happens with computer viruses once they’re unleashed, there’s no need for the human brain in any transaction that takes place.

      As you’ve admitted, I don’t think you’ve characterized the difference between 2nd and 3rd replicators very well. That’s understandable, because the problem is very difficult for people who are not used to thinking of evolution as abstract algorithms.

      Here is the way I understand it — and please correct me if I’m wrong, because I would hate to misattribute my own nonsense to you:


      For memes, the “venue” is the human brain, as well as artifacts in the world (like books and other media) that serve as “batteries” for meme information.

      The fitness function for memes involves selection by humans (usefulness, beauty, parasitism/symbiosis with existing memes, etc).

      Variation arises by virtue of messy transmission and by human-directed invention (avoiding getting into “free will” here). Memes may not copy with perfect fidelity, but our brains often “fill in the gaps” using predictions and expectations — or simply get them wrong, and then they either change or fail.


      The “venue” of the teme is the computer/machine and its attendant storage and transmission vectors.

      The fitness function for temes involves selection by computers/machines (criteria: usefulness, parasitism/symbiosis, others?)

      Temes will almost always copy with perfect fidelity, so the dominant mechanism by which they vary is an unknown — a computer program with errors will simply not run. But if we stopped there we’d be truly lacking in imagination; it’s not hard to envision machines, self-improved by their own temes, coming up with viable mutations both by choice and by virtue of error-correction mechanisms.

      The take-home message from the above that many seem to have totally missed: Temes exist on a plane of evolution that totally obviates the involvement of human brains or any other sort of organic matter.

      Perhaps it would help to make a chart that makes plain the venues and mechanisms of each replicator.

  • This conversation, while ending here, continues on Facebook. Join us there by logging in to your Facebook account and proceeding to our group: http://bit.ly/OnTheHumanFacebook.

  • [...] She has an essay over at the New York Times, The Third Replicator, and will also be engaged in debate with other folks at On the Human, the online project of the National Humanities Center. Her entire essay is reprinted there, just called Temes: An Emerging Third Replicator. [...]

  • [...] Of course cars and books are passive lumps of metal, paper and ink. They cannot copy, let alone vary and select information themselves. So could any of our modern meme products take the step our hominid ancestors did long ago and begin a new kind of copying? Yes. They could and they are. Our computers, all linked up through the Internet, are beginning to carry out all three of the critical processes required for a new evolutionary process to take off. Read the full story… View comments and Sue’s responses here [...]

  • [...] interesting, perhaps, is the ongoing discussion and response to this article: link This entry was posted in Uncategorized. Bookmark the permalink. ← For [...]