Memes and the Singularity
Susan Blackmore is a memeticist – that is she studies memes. The term “meme” was coined by Richard Dawkins in his 1976 book, The Selfish Gene, and refers to a unit of cultural evolution, just as a gene is a unit of biological evolution. Genes and memes are both replicators – information subject to evolutionary processes. Now Blackmore is arguing that we are on the brink of developing a new third replicator – a technological one.
The idea is interesting, and I wonder if Blackmore is aware of the degree to which her ideas mirror those of Ray Kurzweil and his “singularity”. For background, Kurzweil argues that any system that encodes information is likely to form a positive feedback loop – information build on information in an accelerating process.
Actually, this goes back even before life arose on earth. The prebiotic earth saw chemical evolution, which is what likely lead to the first self-repilcating systems that can be said to be alive. At this time we can only know about the prebiotic earth by positing plausible prebiotic environements and plausible chemical pathways that could have lead to RNA (the current theory), and then how RNA could have led to the first cells and life. This process of chemical evolution was relatively slow and with limited potential in terms of what it could create.
Once life arose, however, then we were into a new phase of biologial evolution, which accelerated the process of information evolution. Blackmore jumps forward to cultural evolution, but Kurzweil argues that even within biological evolution the process of the information feedback loop accelerated as some life became more complex. For example, the emergence of multicellular life further accelerated information return, as did the evolution of central nervous systems.
But both Blackmore and Kurzweil agree that the jump to the next plane of evolution was the development by humans of complex culture. Blackmore accepts Dawkins definition that memes are the units of information transfer in culture – the second replicators. Cultural evolution is far more rapid than biological evolution – acquired knowledge can be immediately passed down. Kurzweil emphasizes the fact that as culture advances we gather information about how to gather information, and so this acquisition accelerates.
Both Blackmore and Kurzweil also agree that we are on the verge of entering the next plane of evolution – technology that feeds on itself without the direct intervention of humans. Blackmore calls this type of technology the third replicator (although I guess if you count prebiotic chemical evolution it’s the fourth). Kurzweil argues that once artificial intelligence and nanotechnology come into their own, most if not all technology will be information based, and that information will be accelerating at such a rate that it will transform civilization in ways we cannot imagine. He calls this tipping point the singularity.
Blackmore has not yet named this third replicator, but at her TED talk on the topic she threw out the “meme” that the third replicator may be called “temes” – so we would have genes, memes, and temes. She did not mention “temes” in her New Scientist article I referenced above – it seems because she is not ready to settle on this term. But the teme meme, now that it is out there, may spread beyond her control, and people will use it for lack of a better term.
What is happening, argues Blackmore, is that the location of information storage, selection, and reproduction (the components of a replicator) are moving from human brains to information technology. Living organisms are the substrate of genes. Human brains are the substrate of memes. To say that we have a new replicator it is not enough to have information stored outside of human brains – otherwise books would be sufficient. Humans still create, select, and reproduce the information. But now, increasingly, computer technology is not just storing and communicating information, it is creating it, and even selecting it. Computer programs are increasingly independent of humans for such tasks.
She argues:
Or think of Google. It copies information, selects what it needs and puts the selections together in new variations – that’s all three. The temptation is to think that since we designed search engines and other technologies for our own use they must remain subservient to us. But if a new replicator is involved we must think again. Search results go not only to screens for people to look at, but to other programs, commercial applications and even viruses – that’s machines copying information to other machines without the intervention of a human brain.
I agree – but still ultimately without humans the cycle of information is not really complete. Humans provide the context and use of that information. To be clear Blackmore is not arguing we are there now – just that we are close. Where, exactly, is the line, however. Will it take artificial intelligence before the third replicator truly exists as an independent substrate of evolution? What about computer viruses? Can they be said to evolve separate from any interaction with human brains without the need for AI?
I agree with Blackmore and Kurzweil that we are seeing the beginning of an information revolution that will put humanity and civilization on a new plane of technological evolution. I don’t know if it will be as transforming as Kurzweil argues – or to be more precise, that it will be as transforming as quickly as he argues, but it will probably happen eventually. I also think Blackmore is onto something with her third replicator idea, but I think it is a bit fuzzy around the edges – meaning that it will be difficult to define when we have crossed the line. When we are clearly on the other side of it we might be able to look back and say, there was the line. But being in the middle of the transition it is difficult to see.
Specifically, it may be difficult to define when information replication is fully independent of humans. What if, as Kurzweil predicts, we will merge our AI with our own brains. Human brains may never be fully out of the loop. Or what if true AI is much farther in the future than many futurists imagine? Will sophisticated non-AI information technology ever be truly independent of humanity?
I think the core claims that information feeds on itself in an accelerating feedback loop and that this process can rapidly shift into a new plane where the rules change, possibilites expand, and evolution accelerates – are both essentially correct. One thing is for sure – singularity or no, the future is going to be interesting.
For a critique of conventional thinking on memes, please see my paper, Deriving Common Interests from Animal Origins: The Generative Constraints of Global Polity (at http://www.rkey.com/essays/Simon_DCI_02.pdf ).
My argument is that it is a category error to think of memes as distinct, agent-like entities capable of driving cultural change. If we seek a durable way to speak about the elements of cultural propagation, Dawkins’ description of the extended phenotype provides a quite adequate starting place, and no additional substrate need be posited. Memes can better be described as symbolic expressions within the extended phenotype.
Craig – I agree that memes are symbolic and not distinct. I toyed with getting into this on this post, but really didn’t have the time.
But culture is a separate repository of information subject to replication and evolution from biological information – so it does warrant separate consideration, even if memes are not a great analogy of genes.
“Temes” and memes are analogous to each other – but neither can be divided into “units” of information. They scale all the way up and down without any meaningful boundaries, in my opinion.
Steve, certainly culture warrants separate consideration as a separate repository of information. My point is that the conventional usage of the word “meme” is a horrible analogy, leading to wide-scale reification error.
I have as of yet to see anyone present an adequate justification for believing in the existence of memes however they are cashed out, and that includes the work of Blackmore. The analogy to genes seems to fail at a number of levels, not the least of which has to do with the manner in which such things would be produced and how they would be transmitted from individual to individual. An enormous problem centers around the lack of discreteness of concepts in general, as well as their highly mutable nature and dependence upon other concepts within the large framework for their meaning. A host of other problems exist. As such, I’m skeptical that there is any such thing.
What do you think about the concept of numbers?
Better yet, imaginary numbers.
We are the borg. You will be assimilated. Resistance is futile.
Humans have been building tools to build better and new tools since prehistoric times. I just don’t get what Blackmore is going on about; perhaps she doesn’t understand tools?
Jim, I’m not sure what you mean by memes “existing,” as I don’t think the proponents of memes are saying memes exist in the same way that the computer I’m typing from exists. Memes are more of a conceptual way of looking at cultural evolution, like the way microeconomists conceptualize individual economic transactions using the concept of utility. It’s not that utility “exists” in any sense, but it has been shown to be useful as an explanatory device to some degree in that subject. The issue for memes then ought to be whether it is a useful or useless conceptualization, not whether or not memes “exist.”
I mean that I haven’t seen anyone overcome the large number of conceptual problems that plague the notion of a meme such that that notion could be implemented for any useful purpose. Again, one large-scale issue is the lack of discreteness of concepts in general along with their dependence upon conceptual frameworks for their meaning. As the framework changes, so do the concepts within that framework. So, in that case, are these new memes? Did they undergo some kind of non-biological evolution? What’s the mechanism for that? Being that your conceptual framework likely differs in some way from mine, if I gain some new concept after speaking to you, does it make sense to suggest that I now have the same meme as you? How would that work given that, as our frameworks differ, it is almost certainly the case that our concepts, the one I developed after speaking to you, differ as well? Is it useful in any sense to say we have the same memes when the meanings attached to the concepts that instantiate the memes are, in fact, different? And this is hardly the only issue. There is quite a bit of criticism of memetics in philosophy of science. Skepticism that the idea represents anything useful, “that there is any such thing,” is common.
Al, to solve your question for yourself, answer this:
where is ‘running’?
For ‘running’ surely exists, but you might be equally unable to say so in a coherent manner:- much in the same way as your apparent inability to grasp the reality of memes, merely because you are unable to articulate a meaningless and irrelevant conundrum about said entity.
For the fact that ‘running’ is not a thing, but a process, so is a ‘meme’ not a thing, but a process.
Michael Kingsford Gray, I am not sure what you are disagreeing with me about. I already gave a clear analogy with utility in economics. It is a way of looking at microeconomic transactions, just as memes are a way of looking at the spread of culture. It is futile to argue whether these things “exist” or not because no one is under any obligation to conceptualize it this way. IOW, there are alternative ways of thinking about these two subjects that do not require you to conceptualize the notions of either a meme or a util.
If you want to, you can say that under these schemes, memes and utils don’t “exist,” whereas if we are taking them into consideration, then they do “exist.” Another analogy would be if you are a mathematician working out theorems of real numbers, then infinitesimal numbers don’t “exist” (you can even prove their non-existence), but if you are are a mathematician conceptualizing the hyperreals, then infinitesimal numbers do “exist” under that scheme. I hope you can see the futility of arguing the ontology of these things, because it is really not about whether they “exist,” it’s about whether that conceptualization is useful in shedding light on topics like microeconomics, cultural evolution or the structure of mathematical logic, vis-a-vis other althernative ways of conceptualizing these things. IOW, if you want to object to any of these things, do so by pointing out their uselessness or at least inferiorit to an alternative, not by pointing out that something in them doesn’t “exist.”
Here’s the relevant section from my paper Deriving Common Interests from Animal Origins: The Generative Constraints of Global Polity (at http://www.rkey.com/essays/Simon_DCI_02.pdf ).
==========================
Why are so many people so comfortable with the idea of the Singularity? Why do so many people welcome the idea of abandoning their bodies? Here is where the free will debate reenters the conversation. And here is where it links up with the issue of human malleability. Here also is where IR scholars familiar with the agent/structure debates might be able to help clear things up for the Darwinists. The purpose of this section is to launch a challenge to ethologist/evolutionary biologist Richard Dawkins’ well-known definition of a meme… an idea that is able to copy and transmit itself in ways analogous to the copying and transmission of genes (A fuller description will be presented shortly). This challenge is necessary because the current understanding of memes serves to replicate persistently misleading ideas about the exogenous nature of causality in human affairs, and thus deprecates our respect for freedom.
To begin, it is important to recognize that one of Dawkins’ most significant achievements as a biologist has been to crystallize thinking about what might be called the chicken and egg problem of evolutionary science. For Dawkins, eggs use chickens to make more eggs. That is, the analysis of how genes use bodies (phenotypes) to replicate more genes is coupled to an understanding that selective pressures act upon genes rather than individuals or groups. So, instead of being primarily concerned with how the fittest individuals pass on their genes to subsequent generations, evolutionists have come to focus more on how various behaviors among kin serve to propagate genetic variants.
In Dawkins’ elegant formulation, the beast within is the “selfish gene,” hungry for space in the gene pool. The stability of altruistic behaviors among kin, however costly for some individuals (even up to forfeiture of reproductive possibilities and loss of life), can thus be understood in the context of a genotype’s protective strategy for success. Dawkins also holds that the phenotype can extend beyond bodies into artifacts and even other bodies. Ant hills, for example, would be an expression of ant genes, and a host’s disease symptoms resulting from a parasitical, bacterial, or viral intrusion would be an expression of the infecting agent’s genes.
Dawkins’ concept of memes is where the trouble starts. He describes them as units of cultural transmission that “propagate themselves in the meme pool by leaping from brain to brain via a process which, in the broad sense, can be called imitation.” According to his formulation, memes are to ideas as genes are to phenotypes. That is, memes are selfish entities that can use brains to propagate themselves. The word has, in its own catchy meme-like way, taken hold in common parlance. Since a meme is often portrayed as a kind of mental virus, the idea has been enlisted to hype a kind of advertising called viral marketing. The concept is also widely put to use as an all-purpose explanation for transmission and mutation of cultural ideas. Journalist Robert Wright, for one, has joined the meme bandwagon with fervent enthusiasm.
Memes are also a central concept in transhumanist ideology. Proponents start from the assumption that memes are driven to replicate themselves by exploiting available phenotypes and working to evolve better ones. The most aggressive memes, therefore, would find it advantageous to prompt humans to build superior hosts through which those memes could prosper even more. The most successful of those memes would be prepared to switch substrates altogether, discarding human hosts and taking on cyborg form when that better serves their needs. One seemingly positive analogy for this is a butterfly emerging from a caterpillar-spun cocoon. Another is a parasite that develops by devouring the body cavity of its host, and finally drives the host to suicide when it is ready to breed.
* * *
Dawkins’ error boils down to crediting memes with algorithmic force. This presumed endowment manifests itself as a self-propelled, generative dynamism, and a driving, ontological priority. It has inspired legions of meme fanatics, including psychologist Susan Blackmore, who advocates a worldview that subordinates human intentionality and dismisses the existence of axiomatic human freedom. Dawkins’ error is easily redeemed. His edifying conception of the extended phenotype provides a far better way to account for the replication of ideas. In the paragraphs below, I briefly propose a reformulated description of memes that I believe will prove to be more straightforward and productive for the long run.
First of all, a move toward reformulation starts by accepting that there is no evident or pressing need to elevate memes beyond the status of an extended phenotype. The concept of memes, at its core, is redundant. The genes-eye view is already good enough to explain the rise of culture. Since it is possible to speak distinctively about materially-expressed extended phenotypes, that concept can be recruited to speak also about symbolically-expressed ones. In that sense, all the tools and rules we fabricate can be seen as extra-somatic membranes of our physical and intellectual existence. People are not vessels for memes, but for genes. So then would all our artifacts be such vessels… including memes as well. Describing memes in such terms would provide stronger theoretical grounding for them within Dawkins’ own oeuvre, and thus a more integrated description of the extended phenotype itself.
Another virtue of accounting for memes as a symbolically-expressed extended phenotype of the human species is that doing so could provide a sturdy bridge for interdisciplinary discourse. For example, this proposed reframing would allow for a closer accord with the structurationist concepts described earlier, in which cultural artifacts are seen as the simultaneous medium and outcome of social practice. Likewise, a meme’s reformulated ontological status could be shown to match that of any other discernible social practice within Searle’s conception of social reality. This move, therefore, promises to reveal commensurable perspectives across the disciplined discourses of evolutionary science, linguistic philosophy, sociology, and constructivist IR. But achieving such a commensurable understanding would be no simple task. The artifact-as-agent paradigm is quite persistent and may continue to exert an appeal across otherwise sober and scientifically-oriented disciplines. The challenge for the memetic fundamentalists is their willingness to apply Darwinian acid to their own ways of thinking.
This is not to say that the practice of referring to “memes” must end. The idea resonates. Its vernacular understanding is bound to persist. But there are compelling reasons to question and reject the epiphenominality and subordination of human thought enshrined by Dawkins’ approach.
To Craig Simon,
(Note, I have read your exerpt)
Craig I cannot really see the use of reducing memes to extended phenotype, and I struggle to see that that strategy has explanatory value.
Take for example a boring example such as the chorus of a new pop song which is stuck in my head, and I sing out loud infront of a friend, who also aquires it.
If you wish to explain the pop song as an extended phenotype of my own genes I think you have a problem. Sure you can explain the architecture of my brain and what it is about my brain that leads me to get it stuck in my head (clearly a product of my genes, and some environmental influence). But you can not by any stretch derive the song by understanding my genome. And this technique would seem to ignore many of the important features of the song itself. Its pop melody, the lyrics which are similar to other pop songs at the moment, the rhythm which is inspired and in some sense a mutated version of previous pop rhythms. You may try to swallow all of that up by claiming it is part of the environment that my genes in their extended phenotypic way interact with, but I think you have lost the power of explanation. A theory with no explanatory value is a theory unlikely to make predictions, and therefore untestable.
In terms of a memetic explanation. The pop song is stored in my brain, mutated when I get the melody wrong, though may keep most of its features. And it is copied when I sing it to my friend. The reason it does this is because of its catchy features, the melodies, rhythm, lyrics, puns, which cause my brain, evolved to take note of particular things better than others, to remember it, recall it frequently and sing it out loud more frequently than other things I have heard or thought about. We may ask lots of useful questions about why my brain has the preferences and evolved skills it does but this does not detract from the usefulness of looking at the meme itself and seeing what features it has that aid its storage, replication and mutation (with some fidelity). As it does all these things it can be considered an evolving replicator. And if it does all these things it should be credited with algorithmic force.
Craig how would you design a test to distinguish between memetic and extended phenotypic explanations of the catchy pop song?
Dr. Novella,
I’d like to see you take on Dr. Kurzweil’s health nonsense directly either here or on Science-Based Medicine. He seems to be getting a lot of press for predicting nerd rapture (uploading our brains to computers and attaining immortality). He seems to think that in the meantime we need to get massive supplements, chelation therapy, alkaline water, etc. so we can stay alive for this event. We can conveniently buy all this crap from him and his homeopathic buddy, Dr. Grossman. Dr. Hall has mentioned Kurzweil briefly, but I’d like to see a full frontal assault.
Interesting article, but difficult. Obviously (well, to me, obviously), any idea is `real’ but cannot be equated with actually being physically present, it seems to me. It may be fun to talk about for those inclined to philosophize and conjecture about it (`what ifs’ can be fun), but it’s just too eerie for this mere mortal…
“Kurzweil argues that any system that encodes information is likely to form a positive feedback loop – information build on information in an accelerating process.”
This is a fallacy for three reasons. First, information, by definition, requires energy: something has to convert new data and older information into new (and hopefully more meaningful) information. Second, information building on other information suffers from the law of diminishing returns: it takes more and more energy to gain smaller and smaller amounts of information. Third, as information complexity goes up, the amount of energy required to store and maintain the information rises. (Raw storage can be used, but lots of bits are needed. Alternately, information can be encoded, but that takes effort to devise the coding scheme and energy to encode and decode.)
Here’s an example of reason 2 from my field of laboratory medicine:
A patient comes to the Emergency Room with chest pain. A battery of laboratory tests are performed, and the ER doctor turns the lab results into information and concludes that there is a 90% chance that the patient is having a myocardial infarction (MI). He orders a more specific (and costly) test, and the extra data yields better information: a 98% chance that the patient is having an MI. The ER doc orders another expensive test, the result is equivocal, and the revised chance of MI is 94%. Additional data converted into additional information are not helping: they are wasting time and money.
What I find strange is that Kurzweil and others write entire books about information without understanding information’s fundamental costs and limitations. They act as if information is a stable chunk of granite that popped into being with no effort and can be combined with other chunks of granite to make info-structures. It just isn’t so.
those of Ray Kurzweil and his “singularity”
FWIW, I think it’s actually Vernor Vinge’s singularity, isn’t it?!
I can has cheezburger?
I guess I’m a bit late to this party, but this is a damn interesting thread and I must throw in my two cents.
I had never heard of a meme prior to reading this, but based on its description, I don’t think the world has even seen a true meme yet. Google isn’t a meme, it is today’s version of the stone tool.
A better definition of a meme might be the deliberate programming of information into our own genetic code. When future generations are born with worldly knowledge encoded in their genes, those modified genes will be the true memes.
Science is fundamentally the replication of natural processes. As we learn more about these processes (including genetics), a positive feedback loop will occur that will enable humans to replicate them inherently without constructing a tool. In a rather far fetched sci-fi example, mastery of the processes that allow us to transfer solar energy into a laser beam, when added to our genetic code, could result in humans that shoot photosynthesis-powered laser beams from their eyes.
Inevitably, we will be able to replicate the evolution of the brain and nervous system, which will give birth to AI. With that will come the ability to preserve our own consciousness in a more enduring vehicle embued with even more advanced senses.
That is, if we don’t get wiped out by a meteor first.