LogoGatherer, D. G. (1999). Reply to Commentries.
Journal of Memetics - Evolutionary Models of Information Transmission,3.

Reply to the commentaries on my paper: Why the `Thought Contagion' Metaphor is Retarding the Progress of Memetics

Derek Gatherer

Firstly I should like to thank all the contributors for taking the time to read my article and comment upon it. Secondly, I ought to apologise for the lateness of my reply. This is a consequence of my absence in South America for 6 weeks at the end of 1998, and also my abrupt and unexpected departure from academia at the beginning of February. Consequently I am still having problems with library access, and have been unable to do the reading required to address some of the points made by the commentators. Nevertheless, I hope I can use this opportunity to clarify a few ambiguities that have clearly arisen.

David Hales (1999) makes the point that the `intentional stance' is a necessary feature of everyday life. I agree entirely, but I am unsure that this is really an argument against behaviourism in the laboratory or field. Dennett's work, such as the intentional stance, and (especially) his subsequent `disqualification of qualia' are, as Dahlbom (1993) has pointed out, aspects of behaviourism in its modern guise. Behaviourism is not, contra Heylighen (1999), an outmoded school of psychology but one which continues to evolve and adapt to the changing empirical landscape. Dennett's work, including the `intentional stance', is just one important manifestation of that process. The intentional stance is precisely that, a stance. It has much in common with what Paul Marsden has called the `memetic stance'. In both cases we are required to analyse situations `as if' intentions/memes (delete as applicable) were involved, and ask how such assumptions help to explain the behaviour we see. I would say that the intentional stance is best taken in what Hales (1999) calls `its weakest form' as an `instrumentalist device' - although I accept John Wilkins' (1999) caution that I should not rely overly on intrumentalism. Whether or not it also gives the user `enhanced predictive power', as Hales says, is a matter for empirical demonstration. Hales (1999) holds out some hope yet for the Dawkins B model, in which case I suggest a further argument which may be used against it (Houghton's `shopping list' thought experiment, see Houghton 1997, pp. 162-163). I can also point to the most recent reviews from those actually looking for memes in human brains. Their results are not encouraging to the Dawkins B school (Lounasmaa et al 1996, Gall et al. 1998).

Francis Heylighen (1999) correctly points out that a theory-free science would be scarcely scientific. My determination to throw out most of post-Dawkins B memetic theory may seem to give the impression that I am against theory in principle. It is true that I have no novel theory of my own to offer in its place, but I would insist that my own work is couched in the standard (neo-)Darwinian theory which involves identification of variation, replication and selection. There are newer theories which may also be of assistance in the construction of a behaviourist memetics, for instance the recent work of Wallace et al (1996) and Wallace and Wallace (1998). Theoretical constructs are necessary, I agree. Our job is to find new ones that will succeed where older varieties have failed. Like most memeticists, I came to the subject through theoretical speculation, but I now see myself as a `coal-face' empiricist. Coal-face workers are bound to revolt occasionally against bad (theoretical) management. But that does not mean we are opposed to management in principle.

Heylighen (1999) refers to the interesting contagious suicide problem, and states that "what was replicating was not the actual suicide behaviour, but the idea that suicide is a good way out of an apparently unsolvable problem". I have to disagree with this. I fear it makes too many assumptions. Many of the people who succumb to the strange phenomenon of contagious suicide are precisely those who would seem to have the least reason to do so. `Unsolvable problems' are rarely the issue. If they were, the phenomenon would scarcely be so macabre and fascinating, or so interesting a subject for memetics.

Paul Marsden (1999) suggests that the strategy should be a focus for memetic analysis. I am perfectly happy with strategies that can be analysed behaviourally. I have done a little empirical work on lecture attendance as a student learning strategy. In some cases, it seems to work quite well and in other cases does not seem to have any relevance to the postulated goal (see Gatherer and Manning 1998). Experimental design is difficult, and ethical considerations regarding students' careers do rather limit the extent to which the experimenter can manipulate the system. But I agree with Marsden, strategy-oriented memetics may represent the way forward.

Hans-Cees Speel (1999) requests a clarification of my point regarding the meme-host duality. This is an important plank of my thesis, so I ought to try to improve on my original formulation, as follows.

Individuals have genes in the `ordinary language' sense of the word `have'. Molecular genetics has demonstrated that we do indeed have these genes in the same way that we have the grosser parts of our anatomy. The Dawkins B meme is also something which we could have in the same way. If there are replicating information structures in our brain, then the same ordinary language use of the word `have' could apply to Dawkins B memes just as it does to genes. I would submit, however, that there are no such things. We could have Dawkins B memes, but we don't because there aren't any. Therefore, in ordinary language usage, we do not have Dawkins B memes.

What about Dawkins A memes? Well, these are a very different and much more heterogeneous class of entities, being behaviours and artefacts and all the other messy things that empirical work throws at us. Although there is another ordinary language use of the word `have' which applies to artefacts, for instance I may have a remaindered copy of `Thought Contagion', this is not the same sense in which I have my genes, or in which I could have the fictitious Dawkins B memes.

I speculate that there is perhaps a sense of unease among our non-Anglophone colleagues, concerning such fussiness about the use of words. It is true that ordinary language philosophy in the spirit of JL Austin is not something that has travelled well outside of the Anglo-American world. That honorary Anglo-American, Ludwig Wittgenstein, described ordinary language philosophy as a way of showing the fly out of the fly bottle. I maintain that the meme-host duality is a fly bottle in which the memetics movement has become trapped. Our escape from this particular trap also leads to the demonstration that we can have no true population memetics, as follows.

Population genetics is based conceptually on the assignment of genes to individuals, usually two alleles per locus in a diploid Mendelian population. Those individuals have those genes. They have them in the strong ordinary language sense of `have'. This is crucial. If they ceased to have them in such a sense, population genetics would crumble; all the equations depend on the assignment of genes to individual members of the population. In order to have a population memetics, it would be necessary to assign Dawkins B memes to individuals with the same degree of rigour. But we can't. There are two reasons why not. The first is the trivial one, which is that there are no such things as Dawkins B memes. The second is exemplified in my `Windsor knot' thought experiment, and also in Houghton's (1997) `shopping list' thought experiment.

All we are left with is Dawkins A memes. These are infuriatingly transient entities which fail to satisfy the strong sense of the word `have' which is necessary to a population science, be it population genetics or population memetics. Heylighen (1999) makes some additional point that genes can have multiple phenotypic effects, such as a feather gene whatever that may be, producing several hundred feathers. I do not think this is relevant. A feather gene may produce several hundred or thousand feathers, but they are all on the same individual. That individual has that feather gene, in the strong sense of the word. There is a definite relationship between the feather allele and the individual hosting it. How many feathers that gene produces on that individual is of no consequence to my argument.

This brings me on to another point, made by John Wilkins (1999) which is the hypothetical entity (HYPE) nature of the Dawkins B meme. HYPEs are often discovered to be real entities; that which begins as a convenient theoretical construct can often turn out to have a genuine reality. For instance in the pre-Watson/Crick era, genes were HYPEs. However, there is an important sense in which pre-Watson/Crick genes (`classical' genes we might say) were much less HYPE than Dawkins B memes. Wilkins alludes to `Mendelian genes, observed by their effects'. They are of course observed by their effects, but not just by their effects. I well remember from my student days the long list of sceptical criteria which classical geneticists (my experience was with the Aspergillus school in Glasgow) applied to any claim for the discovery of a novel gene. A classical gene has to be true breeding, it has to demonstrate segregation in crosses, it has to consistently assort independently from, or map together with, other loci etc. The classical gene was not merely a HYPE lying behind the empirical reality of phenotype, but an empirical phenomenon in its own right, defined by all manner of rigorous criteria. The Dawkins B meme, by contrast, has no such criteria, no rigour of any description. It is simply a HYPE lying behind the empirical reality of behaviour.

Wilkins (1999) raises another interesting question: how, in a behaviourist framework can one distinguish between homology and analogy, eg. between little Charlie' accident and Prince Philip's butler? I cannot pretend to have an answer to this. However, I am consoled by the fact that it is equally difficult to distinguish between these things in genetic evolution, even at the molecular level.

Wilkins (1999) also refers to the informational aspects of memes. I think his argument here has affinities with a thought experiment presented to me by Sue Blackmore and friends at Meme Lab, which runs as follows: if I tell you a joke while I have a hoarse voice, you do not tell the joke to your friends using the same hoarse voice, or even using my normal voice. Rather you tell the joke in your own normal voice. We are able to extract a certain informational content from a speech behaviour which does not involve all aspects of the behaviour.

Wilkins (1999) postulates a realm of `culturally significant transmits' (in his Fig. 2) Having just about removed myself from the influence of Popper, I have to confess I feel a little frisson of fear at this spectre of World 3. Having said that, behaviourists as well, of course, have to come to terms with `memes without a hosting subject' (to paraphrase Popper). Popper lurks behind so much of memetics, but his exact relationship to the discipline is still not completely defined.

I hesitate to address the comments of Aaron Lynch (1999), for fear that I leave myself open to further accusations of wilful misrepresentation. Lynch insinuates that my criticism of his work is religiously motivated. Elsewhere, referring to the change from Dawkins B in my earlier papers to my current behaviuourist stance, he ruminates darkly "Exactly what happened to change his mind is unclear". I should have thought it was very clear - reflection on 20 years of internalist memetics persuaded me it was going nowhere. He also insinuates that I was `recently still unaware of the evolutionary epidemiological ("thought contagion/mind virus") explanation for the prevailing forms of monotheism by natural selection'. Alas , I was aware of them, but they are so vacuous as to be scarcely worthy of comment in a scholarly journal. Since Lynch insists, I deal with them below.

Lynch also repeats his peculiar allegation that I have misrepresented him, drawing attention to an `if' which he claims that I have ignored, thus changing the sense of his argument. However, the `if' in question makes no difference. Lynch's picture of menmons arranged ~PQ~TY etc is quite clearly a stack of memory bits, exactly like a computer RAM. His calculus of mnemon conjugations then deals with individuals as hosts of string of the said bits ~PQ~T~Y etc. Why he should now seek to deny this is unclear. It is in his paper.

Lynch also claims that I have a "casual lack of understanding of the term infinite". What can one say? This is nonsense. Of course I understand the meaning and use of the term infinite. And what is the basis for such a strange allegation? His subsequent sentences are unclear, but he appears to be arguing my case. It is precisely because as he says `it is physically impossible for humans with finite brains to construct an infinite number of dates for Napoleon's death' that his calculus of mnemon instantiations is untenable, as developed in detail in my article.

Even more bizarrely, Lynch refers to my `monotheism'. What monotheism? I am not a monotheist, or indeed a theist of any description. However, this outlandish ad hominem argument does present an opportunity to deal with Lynch's own specific theory concerning monotheism, which is that: `In an ancient society, where people believed in numerous gods, a belief that there is only one God had a competitive propagation advantage - not as genetic information, but as mental information. It caused its hosts to devote all of their meme transmission efforts to just one god-meme, allowing this god-meme to out-propagate competing memes.' (Lynch 1998)

As with much of Lynch's work, it is difficult to know if this is a serious proposition or just some kind of witty self-parody. Unfortunately I must assume the former. If monotheism really requires less propagative effort than polytheism, and will out-compete polytheism on this basis, one must be able to explain why Christianity was the successful Trinitarian derivative of a not particularly successful strictly monotheistic religion (Judaism). Christianity, with its almost incomprehensible concepts of 3-gods-in-1 and gods turning into lunchbox items, rapidly became even more polytheistic with the acquisition of a fourth member of the pantheon (the BVM) and a whole panoply of minor deities (the communion of saints). Monotheism, contra Lynch (1998) does not automatically outcompete polytheism, as the example of Christianity shows. Indeed, the incorporation of Roman polytheism in growing Christianity, taking it away from its monotheistic origins, was a crucial element in its spread. One wonders if Lynch thinks at all about his speculations before committing them to publication.

Similarly we have, concerning baseball `Does internally stored information about numbers of players to recruit explain why baseball out-propagates tennis in the US? You have to read Lynch for yourself'. Rather than analyse such a laughable claim, I shall simply quote Caton (1997), who agrees with me that Lynch "suppresses empirical psychology so that personal speculations may flourish."


Caton, H. (1997). Explaining culture: a naturalistic approach. Politics and the Life Sciences. September 1997:319-323.

Dahlbom, B. (1993). Dennett and his Critics. Oxford: Blackwell.

Gall, C. M., Hess, U. S. and Lynch, G. (1998). Mapping brain networks engaged by, and changed by, learning. Neurobiology of Learning and Memory, 70:14-36.

Gatherer, D. and Manning, F. C. R. (1998). Correlation of examination performance with lecture attendance: a comparative study of first-year biological sciences undergraduates. Biochemical Education, 26:121-124.

Hales, D. (1999). Belief has utility - an intentional stance. Journal of Memetics - Evolutionary Models of Information Transmission, 3. http://cfpm.org/jom-emit/1999/vol3/hales_d.html

Heylighen, F. (1999). The necessity of theoretical constructs: a rebuttal of the behaviourist approach to memetics. Journal of Memetics - Evolutionary Models of Information Transmission, 3. http://cfpm.org/jom-emit/1999/vol3/heylighen_f.html

Houghton, D. (1997). Mental content and external representations. Philosophical Quarterly, 47:159-177.

Lounasmaa, O. V., Hämäläinen, M., Hari, R. and Salmelin, R. (1996). Information processing in the human brain - magnetoencephalographic approach. Proceedings of the National Academy of Science (USA), 93:8809-8815.

Lynch, A. (1998). Memetic monotheism. Skeptic, 6:26-27.

Lynch, A. (1999). Misleading mix of religion and science. Journal of Memetics - Evolutionary Models of Information Transmission, 3. http://cfpm.org/jom-emit/1999/vol3/lynch_a.html

Marsden, P. (1999). A strategy for memetics: memes as strategies. Journal of Memetics - Evolutionary Models of Information Transmission, 3. http://cfpm.org/jom-emit/1999/vol3/marsden_p.html

Speel, H-C. (1999). On memetics and memes as brain entities. Journal of Memetics - Evolutionary Models of Information Transmission, 3. http://cfpm.org/jom-emit/1999/vol3/speel_h-c.html

Wallace, R., Fullilove, M. T. and Flisher, A. J. (1996). Aids, violence and behavioral coding: information theory, risk behavior and dynamic process on core-group sociogeographic networks. Soc. Sci. Med., 43:339-352.

Wallace, R. and Wallace, R. G. (1998). Information theory, scaling laws and the thermodynamics of evolution. Journal of Theoretical Biology, 192:545-559.

Wilkins, J. (1999). Memes ain't just in the head. Journal of Memetics - Evolutionary Models of Information Transmission, 3. http://cfpm.org/jom-emit/1999/vol3/wilkins_j.html

© JoM-EMIT 1999

| Back to Issue 1 Volume 3 | Back to Commentaries on Gatherer|