H.C.A.M. Speel

Research assistant Policy analysis.

Faculty of Technology, Policy and Management

Delft University of Technology

Tel. 00 31 15 2785776

Fax. 00 31 15 2783422

Dawkins (1976) defined memes analogous to genes as replicators, and rightfully so. However, it is not generally mentioned and thus probably often overlooked that memes can also be interactors. Both interaction and replication is needed to explain natural selection (Hull, 1988b). Memes, but also genes are interactors if their direct characteristics count in selective events. Hull (1988a) mentions humans as interactors in memetic evolution, not memes. I argue (Speel, 1997) that if we can and do judge memes by their merits without necessary interference of the physical world, which implies a kind of phemotype or conceptual phenotype, memes should count as interactors. If we take the definitions of Hull and Dawkins seriously we cannot deny this conclusion.

Memes are replicators by definition, in analogy to genes. This is the common denominator in memetics, and it is recognized by most writers on the subject. Dawkins (1976) started the definition of memes in this way, and where memeticists differ in opinion on what should count as phenotype-genotype distinction (see Hull, 1988a for some opinions) and even on what memes are (see Benzon (1998) versus Speel (1998)), they do agree that memes are replicators.

A large part of memetics focuses mainly on the dissemination part of memetics that is directly related to replication. The meme as thought contagion is a well-known view that focuses mainly on memetic processes of dissemination. Aaron Lynch’s Thought Contagion (1996) is perhaps the main work giving the meme such connotations. Of course memes may be rightfully seen as disseminating through human systems, but this view alone cannot account for natural or artificial selection involving memes.

Any theory accounting for adaptations by natural or artificial selection must include processes of interaction (processes such as competition and predation in biology) and processes that account for new variation and possibly the re-combination of old variation. I have argued before (Speel, 1996) that memetic theory such as proposed by Dawkins (1976) does not make it clear if meme-complexes such as religions have adaptations and thus adapt. Worse, it does not provide us with possible mechanisms for such adaptations. The same can be said for the work of Lynch (1988) on religion that also lacks explanations on how religions as meme-plexes (co-adapted meme-complexes) somehow incorporate adaptations, by which some religions become more successfully disseminated than others.

This lack of focus on adaptations and mechanisms perhaps explains why memetic theory has only incorporated that memes are replicators, but not that memes are interactors.

In the following pages I shall argue that memes are by definition replicators, but that in 'internal' selection they can be interactors as well. This view is consistent with the view that RNA selection is possible where there are no translation steps such as from DNA to RNA to enzymes (Calvin, 1997). Hull's (1988a) view that genes can be interactors as well is also consistent with this view.

If memes are taken to be interactors, consequences follow for the discussion on what the phenotype must be in memetic evolution. I shall argue, contrary to Benzon (1998), Hull (1988a) and others, that memetic (sometimes vicarious) selection implies two kinds of phemotypes.

Replicators in biological evolutionary theory were introduced by Dawkins (1976) and have become part of the received view on natural selection. Hull has argued that if the theory of natural selection is to be described fully, interactors are the necessary counterpart of replicators. Hull (1980) applies the following definitions in biological evolutionary theory:

• Replicators, or the 'structural entities that are replicated', pass on their structure directly from generation to generation. In principle replicators can only replicate themselves, but they can also produce entities that interact and this indirectly results in the replication of the replicators (by reproduction). These entities are known as:
• Interactors: 'entities that directly interact as a cohesive whole with their environment in such a way that replication is differential'.

In the paradigm example of evolution by natural selection, a sexual reproducing species, the paradigm replicators are the genes and the paradigm interactors are the organisms. Hull argues that replication or interaction alone cannot account for evolution by natural selection, but that both processes are needed:

• Selection becomes defined as 'a process in which the differential extinction and proliferation of interaction causes the differential perpetuation of the replicators that produced them'.
• Lineage is the third term in the Hull scheme and refers in this case to the historical changing entity called the species. Notice that Hull thus takes selection to be more than just 'weeding out'. For him selection refers to evolution by selection on replicators that form lineages.

Interactors play an essential role in biological evolution theory as counterpart of replicators. Brandon (1988) sees organisms, genes, groups and taxa as interactors in hierarchical levels of interaction. It might be noted that different stadia in reproductive cycles often amount to interaction in different arena's: sperm-cells usually compete for a few or only one egg cell; genes in meiosis can compete for inclusion in the sexual cells (meiosis-distorters) and so on. Beyond this, there is probably also interaction between DNA in chloroplasts and other cell-organelles (Eberhart, 1980). It should be noted that some (Dawkins, 1982) use to prefer the concept 'vehicle' in stead of 'interactor'.

The interactor-lineage-replicator scheme as defined above, can and is used in memetic evolutionary processes with only minor changes to the definitions (Hull, 1988b; Speel, 1997). The question then arises what the interactors should be in memetic evolution, since it is of course clear that the replicators are memes. Hull (1988b) has argued that humans in memetic processes are interactors, although he later mentions that humans are agents of some sort in memetic evolution (1988a). He consequently holds memes not to produce interactors since of course humans are not produced by memes (1988b).

Human individuals can indeed be interactors in memetic evolution, but I have argued before (Speel, 1997) that memes can also be interactors in 'internal' interaction or selective interaction (s-interaction). [footnote: I prefer 'selective-interaction', or 's-interaction' to just 'interaction' because this circumvents the confusion with 'interaction' in social sciences that usually means something like 'individuals reacting in some way to each other'. S-interaction implies a result in differential perpetuation of replicators.] Internal selection is selection inside a mind when someone makes up his mind on whether A or B is better, true, more important, or something equal to such criteria [footnote: Note that this is not the same as a decision to remember or to notice something. It is quite possible to know something, and not to agree with it.]. External interaction is interaction where human individuals are the interactors. In external s-interaction human individuals do not question memes themselves in their minds, but they defend memes against other human individuals that act to discredit the same memes. When humans thus in practice act to defend a meme, the interaction is more directly between humans, and not directly between memes. While this internal and external difference of s-interaction might seem far-fetched it is needed to make memetic terminology suited to describe internal selection. If only humans can be interactors in memetic evolution, which Hull seems to propose, there is simply no way to describe whatever factors inside a brain, or making up a mind are responsible for the selection of ideas. Such factors might be called installed complexes of memes (Dennett, 1991), or a self. I think memetic theory ought to describe the internal selection of proposals, theories etc. which we all witness in our own minds, but that we can also verify in other ways, for example by the study of verbal and other behavior.

However, if memes are replicators, can they also be interactors? Can the same entity that is a replicator be an interactor? Can an entity that is code, geno-or memotype, be part of the pheno- or phemotype?

First of all replication and s-interaction are definitions of process or in other words functional definitions (and not structural definitions). If I use a chair to sit on, it is a seat, when I use it to hit someone it might be called a weapon. In principle words like interactor and replicator are definitions of process. The role an entity plays or has played in a process at a specific time and place is decisive for such a definition.

Memes can thus not be excluded from the interaction roles by definition a priori. In fact genes and RNA are also said to be both interactor and replicator. Hull (1988a) acknowledges that genes can be interactors, and Calvin (1997) mentions research on RNA:

Calvin thus argues that parts of RNA, at least in experiments, can be both interactors and replicators.

Parts of DNA are usually not interactors: either a body of an organism or enzymes are the interacting entities coded for by the DNA. However, in some instances the direct physical characteristics of DNA might be decisive in s-interaction: Eberhart (1980) mentions the hypothesis that the size of DNA molecules might be of importance in DNA selection in cell-organelles. DNA would be s-interactor if there was selection on the form of DNA molecules, the stability of them, for instance in a high-temperature environment and so on. In RNA evolution RNA could be interactor if the chemical groups of the RNA-molecules are involved themselves in processes that are decisive for their survival.

We have seen that replicators are also allowed to be interactors because these concepts are definitions of process. We have also seen that genes and RNA can fulfill both roles. The conclusion must be that there is no reason why memes cannot be both replicators and interactors. Notice that the general evolutionary terminology of geno- and phenotype can be caught in the following definitions:

• An entity is part of the phenotype if it is directly subject to selection.
• An entity is part of the genotype if it contains a kind of code, and is replicated

For memetics we can simply substitute pheno and geno with phemo and memo, and we have the memetic definitions:

• An entity is part of the phemotype if it is directly subject to selection.
• An entity is part of the memotype if it contains a kind of code, and is replicated

This leaves us with a problem: why should human individuals be the only interactors in memetic evolution as Hull (1988a) holds? If theories are tested, or only small parts of them, should not the theories be the s-interactors? Hull does speak of conceptual phenotypes (Hull, 1988a), but of scientists as s-interactors. I think that if characteristics of a theory describing reality are the decisive things that make a theory rejected or not, the theory (or a part of it) is a s-interactor. If the academics proposing the theory would be the interactors, they would be candidates for weeding out. Of course scientists can be s-interactors: they might socially become isolated, not respected, ridiculed or even released from their duties, which can count as a kind of social weeding out. But the weeding out of scientists would be a very different matter from the weeding out of the theories they defend. Ultimately scientists and the theory they defend might be weeded out together, but this need not be of course. I think we need to be able to differentiate between the two selective processes, and the difference between internal and external s-interaction makes this possible.

Memetic evolution largely goes on within people's heads. Most memes never become accepted by the mind they are at some moment being reviewed in and get forgotten or are remembered as not being useful, untrue, and so on. Memes, being descriptions of situations, solutions, problems, opinions and so on can play a role in the behavior of human individuals and organizations. In such a view, memetic evolution inside people's minds is vicarious selection (Campbell, 1974), 'producing future' on behalf of the individual or organization involved. It is true that the behavior individuals and organizations show is part of the phenotype of the individual human or organization. When descriptions of such behavior are vicariously pre-tested in the mind of someone, against a virtual model of the real world, this is a test too. It is virtual in the sense that the test stands for something else, a possible action in the real world, but it is a real test in the sense that something is rejected in someone's mind. Even if memes are internally tested without having an effect on human behaviour, it is still a selection- or weeding-out event. If memes are thus internally selected, they are phemotype (or conceptual phenotype), and thus interactors.

Memes are usually defined as replicators, and rightfully so. However, memes can also be tested inside minds, which makes them interactors. Human individuals can also be interactors in memetic evolution if they 'defend' memes in their struggle to be considered true, useful and so on. If a power struggle in for instance politics results in one party dictating policies, the party (or the party members) was (or were) the interactor(s), not the memes. They too won, but through selection of the party (members).

If memetic theory is to include explanations of why specific memes win from other memes, the focus on only the dissemination and replication of memes must be widened. Interaction of memes and humans must be included amongst other things, to explain both adaptations of meme-plexes (co-adapted meme complexes), and adaptations of human and organizational behavior by means of vicarious memetic evolution.

References