Conceptual Problems in Memetics in a Multi-Level View of Evolution and Behaviour

[ University | Homepage | Research | Links ]

Conceptual Problems in Memetics in a Multi-Level View of Evolution and Behaviour

Robert Clewley

Department of Engineering Mathematics, University of Bristol, United Kingdom.

27 January 1999

Abstract. Since Richard Dawkins coined the term "meme" as "a unit of cultural transmission or a unit of imitation" [Daw] there has been contention over the status of memetics as a viable scientific enterprise, and debate over the assumptions and definitions inherent in memetics. This article aims to clarify the status of a meme and its interactions within a scientific framework at ease with modern views of "emergence" and "complexity". The clarification allows some of the fundamental assumptions of memetics to be critically appraised. Key issues are subsequently raised which must be addressed by those trying to develop descriptive or explanatory models of memetic dynamics at the scale of individuals. One such issue relevant to current forms of these theories is that they may appear to unify the language of social interactions but oversimplify the ontology and dynamics of real cultural processes. For instance, the role of genetic, cognitive and affective predisposition, and environmental factors on social behaviour, is often ignored. Consequent problems for theories of memetic evolution are also discussed, despite an overall endorsement of a behaviourist [Gat] and evolutionary [Wil] approach and a "memetic stance" [Mar].

1. Introduction

In a memetic approach to understanding cultural dynamics it is important to be able to discuss an individual's behavioural act as a discrete event in space and time, which may be observed (or described) and subsequently acted out (or described again) by another individual. The very heart of a theory involving such "social contagion" or "cultural transmission" is the intuition that there is some kind of thing being "passed" between people in the material world, which retains much of its abstract form and cultural meaning in the process. This has manifested itself in several approaches to memetics based on a meme-host duality, and often on something akin to a "mind as software and brain as hardware" duality.

For instance, consider the spread of catch-phrases or particular fashion trends amongst certain social groups, or the strong connection between media coverage of suicides of the rich and famous and a subsequent temporary increase in national suicide rates [Ma2]. At first the spread of a behaviour in these examples appears not to be influenced by the genetic or the neurological make-up of the individuals involved. It would seem that in some sense memetic transmission is an independent system.

Despite being able to observe the apparently self-contained and epidemiological nature of the dynamics of social behaviours in a society, it is still important to be clear about two things: (1)exactly what ontological and functional properties a "meme" can possibly have, and (2)in what sense we can sensibly talk about its "storage" in people's minds (or not), and its "transmission" between them via the material world. The analysis of this issue is the principle subject of this article.

The analysis of this issue proceeds on two fronts. One investigates the degree to which we can claim that all "ideas" in the mind of an individual are really of the same type (the "private sameness of ideas", after Aaron Lynch's discussion of the "sameness" of ideas [Lyn]). The other investigates the degree to which two individuals' "ideas" can be claimed to be similar, (the "public sameness of ideas"). It will be argued that both the public and private aspects of ideas are sufficiently intangible as to be an inadequate basis for micro-scale theories of cultural transmission events between individuals (hereon termed "micro-memetic" theories). Indeed, the usefulness to any memetic theories of mentalist terms such as "idea", "belief" and "attitude" is thrown into doubt, following Derek Gatherer's recent arguments [Gat].

Important consequences of the analysis are then discussed. Namely, that the genetic, cognitive and affective predisposition of socially interacting individuals, and the immediate social and physical environment of social situations, are crucial aspects in understanding cultural evolution. Also, that the interpretation and use of formal memetic models requires greater care, especially those used to base a neo-Darwinian view of memetic evolution.

A major conclusion is the endorsement of a purely behaviourist stance [Gat] in an evolutionary approach to memetics (such as that proposed by John Wilkins [Wil], and which will be termed "macro-memetics" hereon). This would be an example of Paul Marsden's "memetic stance" [Mar].

1.1 The use of methodological reductionism

The conceptual analysis presented below makes several uses of a specific view of scientific methodology described in Clewley [Cle]. The methodology represents an idealisation of any scientific theory and its relationships with other theories. In essence it has many similarities to the methodologies discussed in the "complex systems" and cybernetics literature (for instance, the work of Crutchfield [Cru], Heylighen [Hey], or Hoppensteadt [Hop]). However, its detail is not required for the present purposes, since it also has application in sciences which involve extreme formalism and technical concepts.

It will be useful to note, however, that the methodology is based in scientific instrumentalism (e.g. d'Espagnat [dEs]), and focuses on the use of a methodological, or "non-greedy", form of reductionism. In particular, as a methodology it makes no assumptions of "absolute scientific truth" or "independent physical reality". Theories are seen as syntheses of measurements made by imperfect physical instruments, and the consequent perceptions of form that an observer is capable of. An important aspect of the methodology is to consider ideal scientific "explanation" as requiring a logically coherent hierarchy of theories of a phenomenon. Each theory defines a class of model based on elements and interactions abstracted from a particular scale of observation of that phenomenon.

1.2 The process of conceptual analysis

According to the definitions of Laudan [Lau], a conceptual problem in a theory is internal if there is logical inconsistency or vagueness of its basic categories of analysis, and external if there is a logical incompatibility with another theory that has an overlapping domain of application.

For the purposes of this article, the methodologically reductionist style of analysis undertaken below considers memetics as a candidate "emergent level" above an "underlying level" which will be taken to be cognitive psychology. The traditional models of social psychology and evolutionary psychology will be taken as theories having overlapping domains of application with memetics. In some sense those models all use concepts from the level of individual psychology and the levels that are implied below it (loosely, biological and ecological levels of representation, that fit the physical individual into an environment of other individuals and objects), with which concepts involving the social domain are based.

2. A conceptual analysis of memetic assumptions

2.1 The classification of ideas

Before the "sameness" of ideas can be examined, the status of ideas as objects of scientific enquiry will be reviewed. This will provide the basis on which ideas (and hence memes, when they are used to refer to mental objects of some sort) might be classified, and thus compared.

The issue of the classification of ideas is similar to the classic "symbol grounding problem" [Har] in artificial intelligence and psychology. Such problems rely on a distinction between the interpretation of the elementary objects of a theory as:

Interpretation (a): direct representation of a feature present in an accepted notion of "physical reality", or

Interpretation (b): simplifying, abstract representation of a complex feature present in a less abstract level of representation.

This is relevant to memetics because some sceptics argue that a "meme" can be no more than a convenient metaphor which has no basis in physical reality (i.e., a Dennett "skyhook" [De2]), or that memetics is founded on vitalism and homuncularism. Under Interpretation (a) a meme would have an independent existence of its own, and its form or function would be not be reducible to features of any lower-level model of reality. Ideas and their interactions are presumed to have a neuropsychological substrate, and acts of communication involve a physical mechanism. Clearly Interpretation (a) is easily rejected on these grounds. Of course few memeticists accept this interpretation, and would claim to accept Interpretation (b). Despite this, there can be a more insidious problem in memetic theories. After acknowledging that memes and their interactions must have some kind of structural correspondence to the physical world (probably incomprehensibly complex), the functional constraints placed on memetics from "lower levels" can often be ignored. Certainly the detailed psychological or biological substructure of memetic processes is usually considered irrelevant.

One example of this problem is that the apparently particulate nature of memes and their interactions is merely an artefact of the memetic models popular at this time. As models, they only represent some observed aspects of reality at an "emergent" level - a set of possible idealised interactions between abstract entities. Both the interactions and the entities are really much more complex. The instrumentalist methodology typical of the physical sciences would assume that they ultimately map to the behaviour of "physical" systems which do not possess any intrinsic macroscopic boundaries.

For instance, consider a memetic formalisation due to Lynch [Ly2]. The basic objects of his model are atomic or composite "mnemons" (memory abstractions) and human "hosts", and the basic interaction between a host and a meme is a "replication event". Lynch aims for a "non-metaphoric memetics" by avoiding explicit analogy between memes and the replicators found in biology (genes, viruses) or computer science (software viruses). This is achieved using a symbolic calculus to describe the replication and combination processes. This allows a clear description of an abstract mechanism, avoiding reference to the mechanisms of the analogous domains, and so this aspect of the model is internally logically consistent. However, the clarity of the syntactic rules in a formalisation hides the vagueness of the interpretation of the symbols themselves. Formalisation alone cannot rid these basic representations of their intrinsic vagueness, and thus the account suffers from an internal conceptual problem.

2.2 The basis of comparison between ideas

The above arguments demonstrate that memetic theories can mistake the syntactic nature of a formal meme (as an internal mental entity represented in a model) for a semantic nature. To compound this, a lack of absolute uniformity in even the metaphorical interpretation will be demonstrated, meaning that even the syntactic viewpoint has consistency problems. From this it will be shown that the "sameness of public and private ideas" tacitly assumed in many theories is fallacious.

The requisite interpretation for a properly grounded formal model can be seen as having two parts, in order to be fully consistent with Interpretation (b). Each basic concept in a memetic theory-such as a meme (or mnemon), a host, and a transmission mechanism-should have a concrete interpretation in terms of entities at a lower level of representation, and a uniform interpretation in all situations to which the theory is applicable.

The uniform nature of an interpretation is necessary so that all memes have the same underlying form, and thus they can all belong to the same type in their calculus (i.e., a syntactic class). This feature is essential in order that the calculus can apply consistently to all memes. If a uniform and concrete interpretation of a meme is possible then a meme would be a true unit of culture. Here, the intended sense of a meme as a unit is that all ideas would have the same underlying form that allows them to be compared in some precise manner. This is in the same sense that a kilogram is an objectively definable property of all masses, and so as a unit of mass it provides a precise dimensioned quantity on which to base measurements of one kind of "similarity" between physical objects. (It is also in this sense that a gene is an objective unit of heredity.)

2.3 The "sameness" of private ideas

Normally, definitions of a meme in memetic theories enable a meme to have an internal hierarchical structure made up of memes. This means that it is appropriate to attribute a recursive type structure to a formal meme. For instance, a symphony can be interpreted as a meme, but so can the individual tunes within it. This property of a recursive type structure in the syntactic view of a meme has an appealing sense of symmetry, but there is a limitation to its relevance in real situations. Namely, that there can be no exhaustive analysis of the depth of an idea's internal structure in the mind. This is now demonstrated for ideas held by a single individual - so-called "private" ideas.

A private idea is a subjective entity in the mind of an individual, and objectively it has no directly observable properties. In daily life people name, categorise and agree on ideas. These activities may lead to a belief that the concept of a private idea is a tangible, discrete, and decomposable entity. An individual who tries to decompose a private idea into its constituent parts in order to discover its origin in other ideas (related either logically or temporally) will realise that such a belief is fallacious. Although a single, relatively unambiguous, decomposition of an idea is often possible (e.g. a "restaurant script" [Sha]), at every further recursion an unambiguous decomposition becomes more difficult, and the origins of the idea become ever more intangible. An idea is not reducible to a closed hierarchy of constituent ideas because each level of the hierarchy (as an idea itself) suffers the problem of ambiguity in its definition by that individual. With no fixed grounding for the structure of an idea it cannot have a concrete property. With no fixed depth or form of the hierarchy there is no apparent uniformity in the structure of an idea: an idea does not simply correspond to a recursive type structure in a formal calculus.

There is another strong reason against the possibility of uniformity in the "meme as a unit" scenario. Any uniform grounding would have to embrace the disanalogy of the software-hardware characterisation of mind and brain - a dualistic treatment of mind as a causally separate domain. This may, in part, be due to memetic principles drawing inspiration from the nature of computer viruses. The problem is that the role of cognitive and affective factors in behaviour is trivialised by using the epidemiological analogy that the mind is a passive vehicle which merely responds under the stimuli of memes as independent actors.

Here, the "software" in question is the mechanism of memes as typically described by a symbolic calculus, and the "hardware" underlying its form would be the brain processes associated with cognition. A memetic theory is about mental events that represent the parts of the "life-cycle" of memes ("infection", induced symptomatic behaviours including those responsible for "re-transmission"). Commonly, the course of these events are assumed only to be influenced by other mental events. For instance, memetic theories typically consider limited characterisations of emotional events as brands of mental event, but their actual biological basis-as brain processes that are not consciously accessible-is assumed irrelevant, and not considered. Implicitly, mental events are given the same status in many memetic theories, as if they had the same origin in the brain.

It is true that phenomenologically all ideas are mental entities, but their causal origins (and to some extent even the physical origins) in the brain can be argued to vary greatly depending on their content. This is partly due to the lessons of cognitive and evolutionary psychology, and partly due to what is known of the neurological basis for motivation, emotion, intention, inference, and so on. It has been strongly argued in the literature of cognitive neuroscience that there is a richly varied biological basis to the subjective experience of "ideas in general" which should not be dismissed (for instance, see Lister and Weingarter [Lis]). Hence the mental structures underlying different "memes" may work on very different principles and have different sources in the brain. This makes it crucial to know how the abstract forms that these structures appear to generate actually interact at the "hardware" level. Memetic theories tend to homogenise the nature of ideas without a solid justification.

The conclusion is that a real "idea", or at least the subjective experiences we refer to as involving "ideas", cannot be identical to a meme as it is typically described in memetic theories. (Clearly, a meme might be an approximation to an idea in the abstract and unattainable limit of perfectly discrete and decomposable ideas.) This represents an unavoidable minefield for memetic theories that must be carefully traversed if they are to gain credibility among the neighbouring disciplines.

2.4 The "sameness" of public ideas

Two examples of the external type of conceptual problem are possible, and demonstrate the lack of "sameness" between ideas supposedly shared in a social group. The first example (Section2.4.1) involves other theories of social cognition that use cultural transmission as their primary mechanism for cultural variation, but take into account a wider range of psychological factors in its action. This leads to predictions and posited mechanisms which are incompatible with a naïve approach to memetics. The second example (Section2.4.2) concerns a theory of culture which posits a mechanism underlying cultural variation that is totally different in nature to the mechanisms proposed in naïve memetic theories.

Memetic theories can appeal to an intuitive "sameness of public ideas" in order to justify the notion that ideas can be effectively shared in a society of like-minded individuals, and thus take on an objective character. When two people supposedly have the "same" idea in their minds, "we do not use 'sameness' to mean equality in every concrete detail" [Ly2], because as Dawkins [Daw] points out, "almost any statement about two people agreeing with each other would be meaningless." This would appear reasonable, except that one has to be careful about the actual nature of an agreement. For the present purposes, let an idea be characterised as a mental "schema" consisting of a mixture of beliefs and possible behaviours that are relevant and appropriate to some particular situation. An agreement that two people hold the "same" idea requires each party to actively check the contents of the other's schema. As was remarked earlier, no exhaustive or unambiguous definition of an idea is possible, even by the individual who holds it. Furthermore, it is rarely socially appropriate to spend the requisite amount of time necessary to fully discuss the details of each party's version of the idea. Two people may honestly believe that they have reached an agreement on an issue. On the basis of their subsequent actions it can transpire that they actually had quite different ideas in mind at the time. In other words, an agreement guarantees only a degree of similarity between two people's ideas - depending on the time spent discussing the issue and the ability of those two people to successfully communicate their meaning.

To various degrees the subjective and intangible nature of ideas, in their private and public form, is recognised in introductions to many memetic theories. Despite this it seems not to be considered problematic to the validity of the assumptions of objectivity made subsequently. For instance, it is clear that there will be unmeasurable differences in the connotations and interpretations of a supposed "public" idea held by each individual of a group, for the reasons given above. In time, these differences may result in social behaviour that is inconsistent with a model that presumed the idealised transmission of ideas. An exception to memetic predictions of this kind would have occurred because memetic models tend not to account for low-variance differences between individuals. This clearly means that the intrinsically subjective nature of ideas can be problematic to both micro- and macro-memetic approaches.

For memetic transmission mechanisms, such as the process of imitation, there is not even a process of agreement involved between the imitator and the imitated. The detailed mental form and intention of an idea that was merely expressed by some limited physical action is not directly conveyed to an observer: there is no opportunity to verify that an idea formed about the observed actions of a person is anything like the observed person's idea.

The crucial aspect to social interaction is that an observer is always inferring meaning from the very limited direct perceptions that are available. Such inferential processes are analysed in depth in the literature of social cognition theories, and are discussed briefly below.

2.4.1 Social cognition theories

It has been argued that ideas are not "software" that have a form independent of the "host hardware", and are neither discrete nor finitely-decomposable objects. The memetic "host" concept dubiously casts a human mind (a)as a passive receptacle for memes; and (b)as an oversimplistic vehicle for individual memes to control, and (c)as a meme duplicator. In contrast to this passive view of the mind, theories of social cognition claim that social interaction involves a reconstructive process rooted deeply in both the prior knowledge of the individuals concerned and the peculiarities of each situation. As Dan Sperber [Spe, p.31] puts it, "... recall is not storage in reverse, and comprehension is not expression in reverse." As part of the reconstructive process of perception, and as part of the reflective thoughts and actions that follow, a useful first approximation to gross mental structure is the concept of a set of "schemas". This concept originates in social and cognitive psychology. In rejecting a passive view of the mind as a host the presence of a mental homunculus is not being implied. The mind may still be seen as the collective behaviour of a set of interacting, competing, evolving schemas, in a mechanistic fashion. The active view of mind is intended to refer to the role that many of these schemas have in determining the creation, maintenance, modification, and use of other schemas. The dynamics occur between the schemas already present in a mind, as they process incoming sensory signals and their own interactions. It is merely a convenient shorthand to talk of a meme's action to "infect a host".

Fiske and Taylor [Fis] define a schema as a cognitive structure containing knowledge, where the knowledge consists of attributes of concepts and the inter-relationships of those attributes. Their purpose is to generalise and to put structure into the perceptions, inferences and actions relating to recurring social situations. They reflect the rich domain-specific cognitive structures present in the mind. Perceptual and inferential schemas are related to the traditional cognitive psychology concept of "expectation", and action schemas to the concept of a "script". In this article, schema theory is seen as a refined description of the organisation of thoughts, perceptions and actions, compared to the crude concepts of "idea", "belief", or "action". The theory of these types of schema is developed in the above reference.

2.4.2 Evolutionary psychology

It is argued by proponents of evolutionary psychology that cultural evolution cannot be neatly separated from content-specific cognitive faculties that have some genetic basis. Many current treatments of memetics interpret such faculties in terms of a transmitted form of culture, or do not recognise them at all. Whilst it is true in principle that "there is no reason why success in a meme should have any connection whatever with genetic success" [Daw], memetics cannot account for all cultural phenomena on the assumption that memes are entirely "independent replicators" (Cf. Blackmore [Bla]).

For instance, Tooby and Cosmides [Too] provide a strong argument for the existence of cognitive adaptations for social interactions that have arisen by virtue of genetic evolution. As somewhat "innate" abilities they would therefore not be acquired by social transmission. In the same paper, the authors discuss the concept of so-called "evoked" forms of culture. The brief review of their work in this section follows their line of argument closely.

Intuitively, the principle of evoked culture is that a new-born child's mind is not a blank slate on which society chalks up its beliefs and customs. A simple illustration is found in the reference. It involves the types of reciprocity in food sharing that are observed in social groups, where "... a contextual variable-the presence or absence of 'luck' in food acquisition-appears to activate different decision rules governing food sharing" (p.165).

Tooby and Cosmides suggest the existence of highly content-specific evolved mechanisms to account for Social Exchange algorithms [Cos] or, for instance, the observed systems of food sharing. Whilst it is apparent from the literature that this degree of genetic determinism is a contentious issue, their experimental results (and those of researchers in evolutionary psychology) clearly indicate that there is at least some non-general process behind a person's reasoning ability in various social situations. For instance, the differences in food sharing behaviour between a set of cultures studied corresponded to the variance of food acquisition determined by the local ecology [Too]. The view that there is some content-specific process behind social exchanges and the development of particular food-sharing algorithms is the most simple and plausible explanation for cross-cultural similarities in this cognitive domain.

For many other social situations there is also an argument for content-specific, evolved psychological mechanisms which constitute the building blocks of the relevant cultural behaviour. These mechanisms would have evolved to process information about ancient and important adaptive problems, such as language acquisition, sexual jealousy, kin recognition, emotion recognition, and parenting.

The argument behind this view of evolutionary psychologists runs as follows. The standard "socially constructed" view of culture is that the phenomenon of social exchange is culture-specific and historically contingent. Moreover, the standard model says that whenever social exchange is found to exist, it has to be taught or communicated from scratch. In other words, its every structural feature must be specified by the social environment. However, "it is telling that it is just this explicitness that is usually lacking in social life" [Too, p.162]. As discussed earlier, shared assumptions about the world are crucial to the possibility of communication between individuals. This includes the communication responsible for an individual's learning of a culture, either as an immigrant or as a new-born child. Innate domain-specific reasoning procedures supply what is missing in traditional accounts of cultural acquisition.

Consequently, it appears certain that there are at least weakly content-specific cognitive mechanisms that have genetically evolved. These might be seen as "seeds" around which various types of complex, highly content-specific, social behaviour can "crystallise", under the effect of different environmental stimuli. In some sense, therefore, social exchange algorithms might be viewed as a form of evoked culture.

These ideas might be developed more rigorously in terms of "attractors" for "socio-physical schemas" [Cl2]. Briefly, they can be illustrated with a simple example due to Sperber [Spe]. He speculates that humans "have an innate disposition to develop concepts according to certain schemas." His example of why this is an appealing possibility runs roughly as follows. Consider the problems of ostension for a mother who points out and names the species of a bird that is singing in a tree to her infant child. How does the child know what precisely is being given a name: the name could refer to all trees containing birds, or all small, noisy objects, or of that particular bird, or of the underside of its belly? To avoid ambiguity, the child needs some low-level schemas, perhaps reflecting the nature of taxonomy and the economy of expression, which act as "attractors" for the acquisition of new higher-level schemas. These aspects might allow a child to surmise firstly that the mother is referring to the bird in itself, and not as part of its relation to this particular tree, or the fact that the bird happens to be singing. Secondly, these aspects should allow the child to realise that it is the "whole" object of the bird that is being referred to, rather than, say, only its underside. While the child has not yet developed a detailed knowledge of birds and their general relations to other very basic categories in the world, he or she is unlikely to expect the mother to be referring to detailed aspects of a bird. How are these schemas to be acquired socially for such a young infant if they are not somewhat genetically determined?

3. The consequences for memetics

3.1 Two general conclusions

3.1.1 The "forgotten" importance of context

Section 2.4.1 illustrated that inference is a crucial aspect of the process of social cognition. Successful inference depends on an appropriate and detailed frame of reference. This amounts to a reminder of the importance of the personal psychological history of the individuals involved in social exchanges, which therefore include instances of cultural transmission. Furthermore, inference relies on various physical properties of the environment which are relevant to specific social behaviour (i.e., specific events in time and space). Together, it is clear that knowledge of a historically contingent psychological, social and physical context is vital in the interpretation of a social exchange (hereon, the "cultural ecology" of a situation).

Additionally, Section 2.4.2 augments a cultural ecology with a genetically contingent aspect. This might predispose or restrict the occurrence of certain social behaviours in individuals, or furnish their underlying structure, depending on influences from the cultural ecology and the instantaneous form of a social exchange as it progresses.

In terms of developing a memetic theory of cultural dynamics, two individuals' behavioural acts may only be regarded as "similar" by a participant or an observer who can interpret the social meaning and significance of those acts. This is only possible with an understanding of the cultural ecology. Note that the participants are generally not aware of the genetic influences on their behaviour. Since there tends to be much genetic similarity between socially interacting individuals, an awareness of the disparity in the genetic influences is unlikely to be necessary for them! However, the genetic context will be important for the interpretation of social behaviours of a population from the perspective of an anthropologist.

The contextual influences are hard to reconcile with a naïve micro-memetic model, without simply adding primitive mechanisms to the list of assumptions - mechanisms that permit memes to somehow induce behaviours, for instance. Any intentionality given to memes is a weakly-founded attempt to patch over the exceptional cases to standard memetic predictions. It is counter-productive and ontologically unfounded to view the ideas (in the form of memes) as inherently possessing any inductive abilities or other quantifiable properties (such as "fitness"). Such occurrences of this in the past have prompted accusations of mysticism and vitalism. As Susan Blackmore suggests [Bla], a declaration of the intended meaning of so-called "short-hand statements"-which might otherwise be construed as allowing memes intentionality, etc.-is necessary to avoid the traps laid by using imprecise terms.

3.1.2 The use of formalisation in memetic models

In justification of the use of a simplistic abstract model in a micro-memetic model, Lynch claimed: "[Since] science has achieved no direct observation of the neural encoding of ideas, which might have provided a precise language for discussing ideas ... we must settle for an abstract representation of the information stored." [Ly2]

The admission that the mnemon, a basic concept in his model, has no direct interpretation in a lower-level theory of culture or cognition means that it has no precise semantics in a sound, methodologically reductionist, account of behaviour. Unfortunately, what meaning remains is largely due to the analogy with a gene, and for memetics to mature this is clearly undesirable. Furthermore, with only a metaphorical interpretation there is no concrete object which may be precisely identified in the real world. It has also been shown that there would be subsequent difficulty in observing the "life-cycle" of mnemons in the real world due to the problems of public and private "sameness" of ideas.

Some hope comes from realising that whilst the first part of Lynch's claim is undoubtedly true, it does not imply that science will never develop a "language based on a concrete mechanism of information storage" [Ly2]. A systematic attempt to develop such a language is vital so that a purely abstract representation of a meme may only be a temporary necessity. From that a micro-memetic theory might one day emerge that could be directly applied in experimental work.

3.2 The consequences for theories of cultural evolution

The examination of the consequences for theories of cultural evolution expands upon the two general conclusions drawn in the previous section. That is, it develops the case for (1)re-evaluating the significance of context in describing and predicting cultural dynamics, and (2)taking greater care when using formal models to describe complex natural systems such as culture.

3.2.1 Neo-Darwinian memetics

Memetic theories use an essentially Darwinian view of "cultural evolution". In some sense, the process of memetic transmission is Lamarckian [Bla]. That is, an act of imitation consists of a person A copying the behaviour of another person B, not the instructions which B's brain holds for producing that behaviour. Often, however, memetics goes for a stronger correspondence with genetic evolution: neo-Darwinism. This non-Lamarckian interpretation of Darwinism arises from the specious elevation of a meme to an ontological unit, which can then be used as the object of memetic "heredity". The three principles of a neo-Darwinian theory are (a)the source of variation, and (b)the mechanism of selective retention of (c)a particulate heritable entity. In a memetic context, "heritable" refers to any kind of conservative mechanism by which the entities are replicated and "passed on", or "transmitted" between individuals. These principles will now be dealt with separately.

3.2.2 Cultural transmission events as "inheritance"

It has already been argued that a meme is not a concrete "unit" of culture with a uniform interpretation. Since social interaction is an inferential and reconstructive process, the norm is that acts of communication transform ideas rather than simply replicate them. As a result there is no straightforward interpretation of a meme as a "heritable entity".

A detailed model of memetic transmission is provided by Lynch [Ly2], using differential equations of the "population dynamics" type. These form a purely phenomenological model in discussing rates of "replication events", and so on. As with any other formal model of population dynamics in scientific theories, a concrete interpretation of the variables in those equations is needed before one can make a claim that the equations represent any processes actually observable in nature. The semantic interpretation of a "replication event" relies on the unjustified assumptions of unitary memes and the "public sameness of ideas".

Without a clear interpretation of a meme as a "heritable entity", the principles of variation and selection based on the differential fitness of memes also become difficult to interpret.

3.2.3 The source of variation

Underlying the process of memetic variation is an analogy to the concepts of genetic mutation and recombination. However, a typical memetic view is that alterations occurring during a supposed "replication event" are merely accidents (Cf. genetic mutations), and this has often been described using the concept of "copy fidelity". This view is not entirely consistent with the argument that social interactions are mostly inferential in nature. Also, there can be no direct copying process without a discrete thing to copy.

Usually genetic mutation is considered a statistically unbiased process that causes only "minor" changes to a gene. Loosely speaking, this "normal" kind of mutation yields similar genes which compete because (a)they interact mostly with the same enzymes as each other, and (b)they interact with the same enzymes in only subtly different ways. Despite the lack of a universal definition of the gene unit [Daw, p.28], its sound biochemical substrate at least (a)gives an indication of what the dimensions are to the space of possible changes to a gene; and (b)gives the gene a finitely-decomposable property, in that it is composed of base-pairs which are the limit of resolution for the processes of mutation and recombination.

In the case of memetics the lack of a concrete substrate leads to problems. The acceptance that communication events transform ideas to a potentially arbitrarily-large degree, especially in the case of imitation, means that the assumption that only slight alterations occurring during "replication" is only justified when a common "mental context" can be established between the individuals involved. Also, the transformations may have a large bias according to the schemas used by that individual: in other words, a bias due to the cognitive expectations or emotional dispositions of the individual reconstructing the meaning of a communication. The acceptance of a wide range of factors involved in causing variation to ideas means that there is no clear set of dimensions along which memetic mutation can be defined. Thus the analogy to a mutation breaks down.

Similarly, the lack of (a)a clear set of dimensions along which an idea can be decomposed, and (b)a discrete and concretely-defined notion of an idea, means that the concept of memetic recombination has no precise meaning either.

3.2.4 The process of selective retention

There are two aspects to a naïve view of the selective retention process. One is that a host's memes are seen as "a set of mutually assisting memes which have co-evolved a symbiotic relationship" [Gra], dubbed a meme "complex" or "combination". A complex collectively filters out new memes that the host is exposed to (Cf. Dennett's view of "filters" [Den]). The second aspect is that the "new memes" may themselves be seen as competing for the host (as a meme complex), when the memes in question are in some sense "alleles".

It is easy to presume the possibility of interactions between memes when using formal models that force a uniformity and concrete particulate property onto the concept of a meme. Formal memetic models concern the representational aspect of ideas, often understood in terms of the "functionalism" commonly found in the social sciences. A functional description of a situation may be adequate for some purposes, even perhaps to characterise predictions about measurable properties of a society. However, "as abstract objects, representations have formal properties, and enter into formal relations among themselves. On the other hand, abstract objects do not directly enter into causal relations" [Spe, p.62]. Thus an abstract calculus or population dynamics model of micro-memetics may only provide a clear description of meme interactions when those interactions can be justified from some concrete understanding of the situation. For that reason, no a priori justification for general interaction between memes is possible, because the variety of possible biological and cognitive origins of an idea (both within individuals and between them) is historically contingent and currently impossible to determine experimentally.

The concepts of meme co-operation and competition remain vague in the absence of a uniform and concrete interpretation of a meme. Firstly, there is difficulty in being precise about the appropriate decomposition of observed "memes" for the purposes of defining them as "alleles". This requires knowledge that they would both interact with a meme complex on the same basis. A superficial similarity of type between supposed alleles does not guarantee that they will affect the host differently in a suitably predictable or straightforward fashion. The source of the uncertainty is the individual's complex cognitive and affective makeup, and the immediate physical and social context. Secondly, knowing which of the otherwise-unrelated memes of a host will "interact" in a particular situation to act as a filter, and in what way, is again highly dependent on the specific personal, social, and ecological context.

Of course, there is an appealing sense to the way the concept of "meme complex" suggests a deconstruction of the cognitive and affective makeup of an individual into functional partitions, without a central homunculus. The concept may be made precise in the future with developments in cognitive psychology and neuroscience. In the meantime, the only connection between "ideas" that makes sense is the macro-memetic one: i.e., one which can be inferred from experimentally measured correlations between observable behaviours.

3.2.5 Explaining selective retention using "fitness"

Another aspect to selective retention is the need for a concept of meme "fitness" with which one might like to discuss the success or failure of a meme to infect a host. Any simple notion of a universal "fitness" value that is intrinsic to ideas trivialises the role of context. Indeed, there are few real cases where fitness is merely about "costs" and "benefits", even given a specific cultural ecology. But the problems with the "sameness" of ideas make it hard to decompose fitness accurately beyond, for example, "threat", "reward" or "emotive hook".

In terms of formal models, care must be taken not to mistake descriptive devices for ontological properties: any serious sense of "fitness" for a meme must be purely extrinsic and experimentally determined on the basis that the meme is an observable behaviour. Even then, high variation in a cultural ecology can make the experimental notion of (what is therefore inclusive) fitness highly variable also, and thus still hard to quantify and interpret.

4. Conclusion

A central message emerges from the analysis of the "public and private sameness of ideas": memeticists must be clear about the limits on the domain of applicability of their theories, and be aware of possible exceptions to their basic tenets and artefacts in their models. Even though memetic theories have often avoided direct hijacking of concepts from other disciplines, the ethos of memetics remains a jumble of implicit viewpoints inherited from analogies. The analogies may have inspired the early development of memetic theories, but memetic principles need to be re-founded in their own right. They must become mutually consistent, and also consistent with neighbouring disciplines.

The importance of identifying the context of a particular social situation was also highlighted, and will be a unifying theme of any memetic re-foundation. Knowledge of a "cultural ecology" is essential to ground the significance and meaning of social behaviours between the individuals concerned. It links the social and physical worlds in a way that may eventually base a sound behaviourist understanding of cultural transmission at the level of individuals. Similarly, genetic predispositions (such as "evoked culture") are an integral part of the relevant context of a social exchange. A credible micro-memetic theory may emerge from the use of schema theory as an intermediate level to bridge the neurological and evolutionary treatments of individual and group behaviour [Cl2].

Dan Sperber [Spe, p.31] neatly sums up the need to involve lower level processes in theories of culture: "Memory and communication transform information. Thus, to treat representations, whether mental or public, as material causes among other material causes implies rooting the study of thought and of communication in cognitive psychology." Arguing for the involvement of cognitive and evolutionary psychology in a sound understanding of culture means that the levels of representation inherent in those disciplines must become part of the multi-level picture of culture.

Lastly, the examination of the memetic assumptions lends support for:

(1) Gatherer's suggestion that a behaviourist stance towards memes would be the most appropriate direction for memetics to take [Gat]. The analysis of problems with the "public and private sameness of ideas" is perhaps a more elaborate rendition of Gatherer's conclusion that "the concept of a transmitted belief, as opposed to transmitted information, is highly problematic" [Gat]. In fact, all reference to problematic concepts such as "idea", "belief", and "attitude" in memetics could be avoided.

(2) Wilkins' suggestion that memetics would be best studied from an evolutionary perspective [Wil]. Macro-memetics immediately avoids some of the "private and public sameness of ideas" problems when dealing with the statistical variation in observation and measurement of behaviours in large social groups.

(3) Gatherer's argument that there can be no population memetics: without the requisite "sameness" of public and private ideas, it is meaningless to refer to one meme residing in a set of individuals (or one meme having a set of individuals) [Gat]. Thus at the present time, memes are best considered as observable behaviours, even though there will remain some difficulties in defining discrete behaviours precisely [Cl2].

Acknowledgements

I wish to thank Nick Rose and Paul Marsden for their help in reviewing drafts of this essay.

References

[Bla] Blackmore, S., 1996, Memes, Minds and Selves, paper presented in the London School of Economics seminar series "About Biology", November 28th 1996.

[Cle] Clewley, R. H., 1998, The Role of Scientific Methodology in Demystifying Emergence and Complexity, See http://zeus.bris.ac.uk/~enrhc/research.html

[Cl2] Clewley, R. H., in preparation. See http://zeus.bris.ac.uk/~enrhc/research.html.

[Cos] Cosmides, L., and Tooby, J., 1992, Cognitive Adaptations for Social Exchange, Ch. 3 of "The Adapted Mind: Evolutionary Psychology and the Generation of Culture", eds. J.Barkow, L. Cosmides, J. Tooby, Oxford University Press, New York.

[Cru] Crutchfield, J. P., 1994, Is Anything Ever New? - Considering Emergence, in Integrative Themes, eds. G. Cowan, D. Pines, D. Melzner, Santa Fe Institute Studies in the Sciences of Complexity XIX, Addison-Wesley.

[Daw] Dawkins, R., 1976, The Selfish Gene, Oxford University Press, Oxford.

[Den] Dennett, D., 1993, Consciousness Explained, Penguin Books.

[De2] Dennett, D., 1995, Darwin's Dangerous Idea, Penguin Books.

[dEs] d'Espagnat, B., 1989, Reality and the Physicist, Cambridge University Press, Cambridge, England.

[Fis] Fiske, S., and Taylor, S., 1984, Social Cognition, Addison-Wesley, Reading, Massachusetts.

[Fos] Foster, C. L., 1992, Algorithms, Abstraction and Implementation, Academic Press, London.

[Gat] Gatherer, D., 1998, Why the 'Thought Contagion' Metaphor is Retarding the Progress of Memetics, Journal of Memetics - Evolutionary Models of Information Transmission, 2.

[Gol] Golomb, D., et al., 1992, Clustering in Globally Coupled Phase Oscillators, Physical Review A, Volume 45, Number 6, pp. 3516-3530.

[Gra] Grant, G., 1990, Memetic Lexicon, html document, the Principia Cybernetica Project website, <URL?>.

[Har] Harnad, S., 1990, The Symbol Grounding Problem, Physica D 42, pp. 335-346.

[Hey] Heylighen, F., 1991, Modelling Emergence, in World Futures: the Journal of General Evolution, special issue on Creative Evolution, ed. G. Kampis.

[Hop] Hoppensteadt, F. C. and Izhikevich, E. M., 1997, Weakly Connected Neural Networks, Springer-Verlag, New York.

[Kan] Kanitscheider, B., 1991, Unification as an Epistemological Problem, in Advances in Scientific Philosophy, eds. G. Schurz, G. Dorn, Rodopi B.V., Amsterdam.

[Kup] Küppers, B. O., 1990, Information and the Origin of Life, Massachusetts Institute of Technology Press, Cambridge, Massachusetts.

[Lau] Laudan, L., 1977, Progress and its Problems, Routledge and Kegan Paul, London.

[Lew] Lewontin, R. C., 1978, Adaptation, Scientific American, Vol. 239, No. 3, pp. 156-169.

[Lis] Lister, R. G. and Weingarter, H. J., 1991, Perspectives on Cognitive Neuroscience, Oxford University Press, New York.

[Lyn] Lynch, A., 1996, Thought Contagion: How Belief Spreads Through Society, Basic Books, New York.

[Ly2] Lynch, A., 1998, Units, Events and Dynamics in Memetic Evolution, Journal of Memetics - Evolutionary Models of Information Transmission, 2.

[Mar] Marsden, P., 1998, Memetics and Social Contagion: Two Sides of the Same Coin?, Journal of Memetics - Evolutionary Models of Information Transmission, 2.

[Ma2] Marsden, P., 1998, Operationalising memetics - Suicide, the Werther effect, and the work of David P. Phillips, Proceedings of the 15th International Congress on Cybernetics, Namur, Belgium.

[Mas] Mason, K., 1998, Thoughts as tools: The meme in Daniel Dennett's work, Proceedings of the 15th International Congress on Cybernetics, Namur, Belgium.<URL at PCP needed>

[Sha] Shank, R. C., 1982, Chapter 1 of Dynamic Memory: A theory of reminding and learning in computers and people, Cambridge University Press, Cambridge.

[Spe] Sperber, D.,1996, Explaining Culture: A Naturalistic Approach, Blackwell Publishers, Oxford, England.

[Stö] Stöckler, M., 1991, Reductionism and the New Theories of Self-Organization, in Advances in Scientific Philosophy, eds. G. Schurz, G. Dorn, Rodopi B.V., Amsterdam.

[Too] Tooby, J. and Cosmides, L., 1992, The Psychological Foundations of Culture, Ch. 1 in The Adapted Mind: Evolutionary Psychology and the Generation of Culture, eds. J.Barkow, L.Cosmides, J. Tooby, Oxford University Press, New York.

[Wil] Wilkins, J. S., What's in a Meme? Reflections from the perspective of the history and philosophy of evolutionary biology, Journal of Memetics - Evolutionary Models of Information Transmission, 2.

[ University | Homepage | Research | Links ]

This page was written and maintained by Rob Clewley