Gene-culture Coevolution


Evolving Creative Minds IV

Whereas most animal evolution arises from the differential replication of genetic information, human evolution obligatorily involves the differential transmission of both genetic and cultural information. "Cultural information" is a term of convenience, rather than literal descriptive fidelity here; it refers loosely to such things as the stories, beliefs, and ceremonies shared symbolically among individuals in every human society. Changes in the abundance of a variant of a story or ceremonial behavior over time comprise cultural evolution in its simplest conceivable form. Sociobiology proposes that genes and culture do not evolve independently, on separated, isolated tracks. The neurobiology of human mental development makes them co-dependent, resulting in the process of gene-culture coevolution (Lumsden and Wilson 1981). Gene-culture coevolution in human beings appears to be based on gene-culture transmission, a process of organismic growth and development in which innate learning capacities respond to certain forms or types of cultural information in preference to others, demarcating the central tendencies around which cultural diversity plays. Thus, while human cultures differ hugely in the details of their kinship terminologies, no human culture lacks terms for making sense of one's place in the succession of young and old.

Human history links the branching trees of cultural and biological evolution.

Ed Wilson and I have introduced the specific term "epigenetic rule" to refer to the patterns of genomic expression that chaperone the individual mind's development (Lumsden and Wilson 1981, 1983). Epigenesis is the total process of interaction between genes and environment during the course of organismic development. Each epigenetic rule affecting mind and behavior is approached as comprising one or more elements of a complex sequence of events occurring at various sites throughout the nervous system and mind. We have found it useful to organize these elements into two principal classes: primary epigenetic rules, which range from initial sensory filtering to perception, and secondary epigenetic rules, which mediate the development of central traits and capacities such as temperament, personality, and beliefs, through which people are predisposed to foster the transmission of certain variants of cultural information over others. The primary rules are the more genetically restricted and inflexible of the two classes; cases involving vision, hearing, taste, and smell have been identified. Each class exerts important effects on the mind's capacity for self-organization, and has resulted in parallel or convergent evolution in independently derived cultures.

In gene-culture coevolution a circuit of reciprocity operates: culture is generated and shaped by the biological imperatives embodied in the epigenetic rules while genes shaping the epigenetic rules shift in response to changing cultural opportunities. But while natural selection figures prominently in gene-culture coevolution, as it does in the more familiar kin selection and reciprocal altruism of animal sociobiology, the outcomes of its effect on social populations can be very different from those expected on the basis of genetic evolution alone. Epigenetic rules for culture learning can create a strongly nonlinear couplings between genetic and cultural evolution, with surprising results. The diversity of possible evolutionary outcomes may be comparatively greater, and the speeds with which they are approached by the population higher. Altruistic behavior may spread through a gene-culture population without the aid of kin selection, reciprocal altruism, or any of the mechanisms traditionally envisaged to account for animal social behavior.

Gene-culture coevolution is a causal whirlpool in history, where culture is shaped by biological imperatives and genes shift in response to changing cultural opportunities.

The study of gene-culture coevolution is a development in sociobiology and evolutionary science that is intended to help create a network of explanation between biology and the social sciences. It is designed to include all cultural systems, from the protocultures of chimpanzees and dolphins to the heterarchical, protean cultures of human beings, as well as forms of culture previously conceivable only in the imagination. In pursuing this course, however, sociobiology runs headlong into human creativity as a force in history (Lumsden and Wilson 1981, Findlay and Lumsden 1988, Lumsden 1997, 1998). Our attempts to map the circuit of gene-culture coevolution, as it passes through the imagination on the way from biology to society and back again, has raised puzzling questions about the limits to sociobiological knowledge and a scientific understanding of mind in general.

Horizons of the Unimaginable

One of the crowning achievements of modern science is, surely, the mathematical theory of population biology and evolutionary genetics, disciplines which reveal the quantitative form of the principles governing the ebb and flow of gene variants across time and space (Provine and Debus 1987, and Roughgarden 1996 for exciting surveys). In its population genetic interpretations of human societies, sociobiology has drawn extensively from these principles. In its extensions to gene-culture coevolution, sociobiology has furthermore asked if the connections among genes, mind, and culture are also amenable to expression in the form of mathematical propositions, so that science might be provided with coevolutionary principles possessing quantitative utility. Mathematical sociologists, as well as anthropologists interested in the mathematics and exotic formal logics of human lifeways, have asked similar questions, albeit more often focusing on non-biological approaches to behavior and group organization (e.g. Fararo 1992, Hage and Harary 1996, Kiel and Elliott 1996).

What draws all these pursuits to the goal of mathematical representation is, in part, the notion of a "system", or multi-part object within a domain of scientific discourse, as a thing with time-variable deep attributes called states. To be in state at a time t is to determine fully all relevant knowable attributes of the system at that time. The set of possible states across which the system can range is often referred to as its "state space". and the temporal change of state its "dynamics". Dynamical systems theory, which cognitive science, population genetics, mathematical sociobiology, sociology and anthropology all adapt in one form or another in order to describe temporal change, presumes that the relevant system dynamics can be expressed in the form of mathematical propositions. Hugely successful in the physical sciences, the dynamical systems perspective is a comparative newcomer to biology, psychology, and the social sciences, where its merit continues to stimulate vigorous dispute.

A system in action

Consider for a moment how human gene-culture coevolution might be approached in such terms. Once the possible states of genetic information and cultural information an individual can posses have been specified, these are combined into a set that defines the possible outcomes of conjoint genetic and cultural transmission. This set is the state space of the population. Then, an updating rule, or coevolutionary dynamic, is formulated. This rule operates on probability distributions defined over the state space. The probability distributions specify the frequency with which individuals of a given genetic and cultural pedigree are going to occur in the population. The action of the updating rule is to transform the probability curve describing the population at an initial time t0 into a succession of curves for population structure at all future times t. Our understanding of the population therefore is expressed in terms of the updating rule (the coevolutionary dynamic), the probability distributions, and the state space. In those cases where, due to the complexity of the population or the lack of sufficient data, mathematical description in not practicable, similar general considerations still apply since reasoning about temporal change necessarily involves the identification of attributes bearing more than unary value, the assignment of specific values (whether qualitative or quantitative) to specific time periods (short or long), and the search for pattern in the tempo and mode of the observed changes.

Any attempt to place the imagination and creativity within this kind of framework at once fails because the underlying state space can no longer be considered a constant structure. With each new creative act, each new innovation, new elements are added to the set of cultural information, and existing elements may disappear as new patterns of usage replace old. The state space has changed. Relative to these new cultural configurations, the probability distributions, defined for the original state space, are no longer well defined. Equally troubling, the updating rule expressing the dynamic of gene-culture coevolution must somehow be reorganized to account for the new innovations and their effects on the society.

It is evident, then, that without some way of dealing with the creative behavior, dynamical system treatments can work only when the space of cultural information is not changing, that is, between times that creative acts change the culture. In some hypothetical stone-tool culture of early hominids, the period of time between the event horizons demarcated by successive creative acts might be fairly long - years, or perhaps even millennia (Pfeiffer 1982, Tattersall 1993). But in the human cultures established since historians began their craft, creativity and innovation are frequent occurrences. For modern postindustrial civilization, imagination streams into culture so fast as to be effectively continuous relative to the time scales of biological and macrosocial evolution: a far cry from the isolated, punctuated impacts of innovative change perhaps more typical of Paleolithic times.

Which way to the future?

The material consequences of human imaginative activity, realized through cultural innovations, therefore make the future (and our evolutionary pasts) unpredictable in a manner radically different from that treated by modern theories of nonlinear dynamics and "chaos", where a system might wander around a space of possible states in a highly complex fashion, uncomputable except by impossible automata endowed with infinite numerical precision. Even with chaos, the state space is fixed, a fore-ordained arena for temporal pattern formation. Creative behavior yanks these spaces apart, transforming arenas of change into changing arenas for evolutionary transformation.

Population biology has of course dealt with a similar-sounding problem before, namely the process of genetic mutation and the changes it introduces in units of genetic information as they pass between generations. Genetic mutation events, however, are not goal-directed or intentional activities, and their effects on the individual phenotype and behavior generally are random with respect to the force natural selection exerts on the population (but see Lumsden 1998 for a summary of the recently elaborated concept of directed mutation). One therefore suspects that creativity and the imagination are likely to demand treatments quite different from the random-walk dynamical equations used quantitate evolution in the presence of mutating genes.

One trick, attractive at first sight, is to anticipate what innovations might occur (warp drive or a total cure for cancer by the 23rd century?) and incorporate them into the culture set to begin with. The culture might at first be constrained to move only through pre-innovation parts of its state space, but as the innovations occur its options increase, and more and more of the culture space becomes accessible to the evolutionary process.

While formal methods based on this idea of "soft predeterminism" can help for short runs of simulated history (Lumsden and Wilson 1981, Findlay and Lumsden 1988), they are quite deficient in the long run because they require us to ordain the initially unknowable - what will be discovered, with what consequences - right at the start, when we set up the culture space. The fallacy is supposing that our initial conditions of explanation can be forced hold the answer to the problem we actually want to use them to solve, namely forecasting creativity. Making the dimensions of the state space continuous, and thus allowing a continuous infinitude of possible cultural variations to be portrayed compactly, rather than using a discrete enumeration of culture states as I have, for simplicity, been doing here with my talk of "units", does not solve the problem. The move only obscures it by pushing the locus of disruption onto the dynamical update rules. Nor, as I have noted above, is the issue one of admitting the limits of abstract mathematical models and switching to qualitative modes of analysis and explanation. Talk of change must be talk of change in something, which leads us right back to a literal or metaphorical dependence on states and state spaces.

If we are to keep the creative mind and gene-culture coevolution from slipping entirely beyond the limits of scientific knowledge, a type of theory is needed that keeps the future and past "wide open" - in other words, with room for the initially unimaginable to happen - while remaining, in some appropriate sense, hypotheticodeductive. Is this possible?

Hung by our fixed texts.

It is, so long as we are prepared to take part in a new kind of scientific language game, which cannot be played with fixed texts. Suppose we have a theory of gene-culture coevolution that includes a principled treatment of creative behavior and the innovations consequently introduced into the society. Using the theory, we set up a population model to start at time t0 and predict the pattern of gene-culture coevolution for times t >t0 . Since the creative mind innovates, and innovations can change the cultural and biological milieu of individual development, choice, and group decision making - including those which influence human creativity and the directions it takes - it seems very unlikely that any one set of principles, based on our initial specification of the model at some time t0, will remain valid for very long. Even if these principles do the trick at t0, a key innovation at time t1 (genetic engineering of the neocortex and amygdala, for example) could radically alter creative behavior in the population after time t1. The theory, and all models based on it, would therefore be invalid for times beyond the t1 event horizon unless the model could somehow absorb the consequences of the innovation and adjust itself accordingly. "Adjustment" would entail, minimally, a self-directed production of suitable changes in the model's state space and the gene-culture coevolutionary dynamic that is driven by the creative behavior and its outcomes.

Scientific theories in the form of fixed texts or assemblages of mathematical propositions cannot do this. They are too "dumb" for the job because they lack any apparatus by which changes in their domain of discourse can automatically be cut-and-pasted back into their content. To do so, the theory itself would have to be a kind of dynamic entity, directed toward the goal of revising itself to accommodate the consequences for its subject matter of the events it is, for a while at least, competent to predict. Evolutionary theories of the creative mind must therefore themselves be capable of special types of creative behavior.

As recently as a few years ago, the notion of self-rewriting theories, which cross hypotheticodeductive event horizons by accommodating the unanticipated, would have been just a metatheorist's pipe dream. This is, I think, no longer the case. The advent of the World Wide Web provides the working scholar with a relevant means of composition ("progenitorial authoring", perhaps?): hypertext that can be assembled or written "on the fly" to craft a custom text in real time. To this advance in expressive capability may be added conceptual progress in theoretical computer science, with its improved methods of programming software "agents" that have autonomous, adaptive, goal-directed behavior (McFarland and Bšsser 1993, Maes 1995, Watson 1996). Computer-based agents are user-friendly pieces of software with their own agendas - searching the WEB for your favourite topics, reporting hot stock-market events, or screening your e- mail, for instance - combined with a knack for learning how to do better as they go.

Consider, therefore, the attributes of a scientific theory created not as a fixed text but as a digital agent, a type of computer-based "artificial life" (Levy 1992) designed to adapt itself to the consequences of creativity and innovation as they occur in its model simulations. The theory's software interface allows you to enter the gene-culture population's initial state space and the coevolutionary dynamics, along with the propositions initially applicable to individual creative behavior in the society plus any fine-tuning of the adaptive programming the theory will use to rewrite itself as innovations occur in the simulated population.

Doing theory

Together with the probability curves for the genomic, epigenetic, and cultural organization of the population, the output includes the text of the current valid form of the theory, along (given sufficient disk space) with all its previous versions. A full epistemic deck. A modest but promising example of such capability, in which the theory agent rewrites itself to explain fairly complex patterns of affective events in a simulated individual mind, is the EVA (emotive virtual agent) model being investigated by David Kreindler, Nicholas Woolridge, and myself. I have recently reported on EVA elsewhere (Lumsden, Kreindler, and Woolridge 1998).

My suggestion, then, is that scientific theories authored in the form of adaptive, self-evolving digital agents may help shatter some important limits to evolutionary knowledge - limits that at present obscure the creative mind and its role in the human past and future.

Feeling With, In Conclusion: Evolutionary Theory as Bardic Agent

There is in principle nothing to prevent one from wedding digital agency to character animation. Through the art of computer-generated faces, gesture simulation, and speech synthesis the scientific theory, now instantiated as a digital agent, is also endowed with the illusion of life (Thomas and Johnston 1984, Parke and Waters 1996). When you look at the computer screen, you don't just see probability curves and self-generated revisions of abstract coevolutionary postulates. You also see an animated character, the embodiment of the theory- agent itself, communicating expressively with you about what is going on in the model it is simulating for you, and in itself as it absorbs the implications of the simulation.

The last thing we are likely to need, of course, is theories that take up our hurried careers by lamenting (or crowing) about themselves and their achievements. It is clear, however, that the bardic tradition is an old one in human culture, probably as old as the human mind itself (Pfeiffer 1982). Its modern descendants - cinema, theatre, television - engage us by engaging, in part, our capacity to feel with the other, to satisfy ourselves about what it is to be like the esteemed or loathed one before us.

The face of future theory

It is possible that, as they evolve, theory-agents of the kind I have portrayed in the previous section will begin tapping the rhetorical power of character animation and, in so doing, gain the ability to re-engage our sympathetic imaginations about their subject matter. It will then be up to us to decide if the agony of the atrocities tracked by our poverty- or war- simulations, or the bliss of the simulated creative jumps, should belong in our knowing minds, alongside the numbers and evolutionary statistics. Will science practiced through such bardic theories be better or worse, and the veils surrounding human wisdom any less tangled?

Among academics at least, it has become almost conventional to erect closely guarded barriers between the practice of depicting that which is true about what we behold, and practices which invite us to become that which we behold. In so doing, it seems to me that we have stepped away from modes of comprehension that MH Abrams (1953) noted we in fact desire to use on ourselves, and by so doing illuminate how human nature could be at once spontaneous yet regular, lawful but unlegislated, and rationally explicable after the fact yet vitally intuitive at each moment of personal being. Sociobiology, perhaps partially re-created in striking new metalanguages, has a central place in regaining what we have, for the once-supposed best of reasons, given up.


Research support from the Medical Research Council of Canada, the Natural Sciences and Engineering Research Council of Canada, the Atkinson Charitable Foundation, the J.P. Bickell Foundation, and the Harry Frank Guggenhein Foundation has been instrumental to the work reported here, and is gratefully acknowledged. All images copyright © 1998 PhotoDisc, Inc.


Fehler melden
Telepolis zitieren
Vielen Dank!

Medien im Krieg

Krise zwischen Leitmedien und ihren Rezipienten


Eine kurze Geschichte des Quantencomputers

Wie bizarre Quantenphysik eine neue Technologie erschafft


Mit dem Schalter am linken Rand des Suchfelds lässt sich zwischen der klassischen Suche mit der Heise-Suchmaschine und einer voreingestellten Suche bei Google wählen.


Zum Wechseln zwischen Heise- und Google-Suche

Verlassen und Zurücksetzen des Eingabe-Felds

Buchstaben-Taste F
Direkt zur Suche springen


Mit dem Schalter am linken Rand des Suchfelds lässt sich zwischen der klassischen Suche mit der Heise-Suchmaschine und einer voreingestellten Suche bei Google wählen.