Field Analysis on Competitive Dynamics and Cultural Evolution of Religions and God Concepts
Metanexus Bios. 6,167 Words.
“The sudden emergences of the kingdom of God are like seeing God in the fluids of a waterdrop. You both need to have the curved structure of the fluid drop “out there”, and you need to adjust yourself “internally” to seeing God in that fragment of reality. Nothing goes without the other. For in the world of autopoiesis, no adaptation happens without self-adaptation.”
Field Analysis for a New Research Initiative by the Metanexus Institute on “Competitive Dynamics and Cultural Evolution of Religions and God Concepts”
By Niels Henrik Gregersen, University of Copenhagen
The following field analysis aims to describe some distinctive theoretical approaches that commend themselves for studying the evolution and dynamics of religious systems (including concepts of God and divinity) from a perspective informed by the natural sciences.
The theoretical approaches to be presented are varieties of computational complexity studies: Cellular Automata (CA), Complex Adaptive Systems (CAS), Game Theory (GAT) including Rational Choice Theory (RCT), and Autopoietic Systems Theory (AST). In each case, the theoretical background of the approaches will be presented, and some examples of their application on religious evolution and concepts of God and Ultimate Reality will be indicated.
The purpose of this analysis is to facilitate new research programmes that either apply some already elaborated explanatory models on the empirical case of religious evolution, or develop new science-based methods for dealing with the emergence, evolution and stabilization of religious semantic systems. From the outset it should be admitted, however, that the field of ” Competitive Dynamics and Cultural Evolution of Religions and God Concepts ” is still in its nascence, and is by no means makes up a coherent research field. However, it is a highly promising field that commends itself to further study and calls especially for inventive scholars who are able to develop new methods and approaches, and to use methods known from the biological and economic sciences on exploring the cultural dynamics of religious systems. It should also be noted that the approaches mentioned here are far from exhaustive; other approaches are analysed elsewhere on this website, and these should be consulted as well.
Common to the approaches to be discussed below, however, is the combination of formalistic or computational aspects with a Darwinian perspective on the evolution of cultures. Thus the underlying assumption is twofold. First, it is not only possible but also advantageous to use methods known from the natural sciences in the understanding of the evolution of religious systems of meaning. Second, any cultural and religious semantics has to cope with the problem of reproducing itself, and to adapt itself to new contexts under evolutionary pressures analogous to those known from the fields of biology and economics. The computerization of the sciences since the 1970s indeed offers attractive formalistic approaches to the study of the dynamics of cultural systems. The main scientific question is here not so much “What are the constituents of culture (natural resources, institutions, communities, language etc.)?”. New questions are born, such as “How does nature and culture work?”, and not least, “How do natural and cultural systems evolve?”.
Computational Complexity (CC) theory, however, is an umbrella term for a wide variety of studies on the formation, development, and propagation of patterns, some more general, some arising under specific organizational conditions. The field builds in particular on thermodynamics, information theory, cybernetics and evolutionary biology, but also on economics, systems theory, and other disciplines. Since complexity research consistently crosses the boundaries between the inorganic and the organic, the natural and the cultural, the field is likely to influence the future dialogue between science and religion as two major cultural forces of the 21 st century significantly.
2. Cellular Automata (CAs) and the Cultural Dynamics of Religion
Let me begin with the study of cellular automata (CA). The idea of cellular automata goes back to John von Neumann and the program of cybernetics in the early 1950’s. Cybernetics was concerned with the construction of control systems that are able to move, channel and combine information bits according to pre-described computational rules. For example: If situation A, then do AB; if B, then do BAB.
A Cellular Automaton is a primitive artificial world. Its “space” is a grid consisting of equal squares, usually on a two dimensional lattice. The initial conditions of CAs can be set either as specific or as random states. The “time” of CAs depends on the transition rules that determine how the cubic cells are to be changed, moved, removed or reproduced at each computational step. CAs thus use individual based modeling, that is, the “organisms” are placed in cubic cells on the grid, and their “actions” are specified by the number and features of cells in their immediate neighborhood.
In 1969 John Conway developed an efficient computer model called Game of Life ( see Gardner 1970). The rule for this two-dimensional CA is that the state transitions depend on the states of the other eight neighbors of the cell (also the diagonal ones). The rules are so-called “totalistic” rules in so far as the rule is determined by the total number of the neighboring colors, not on their particular positions relative to the cell. Furthermore, the cells have only two states, black and white. Now the transition rules are as follows:
* If two of the neighboring cells are black, the cell is unaltered (mimicking equilibrium).
* If three are black, the cell becomes back (mimicking reproduction).
* In all other cases the cells become white (mimicking extinction).
Most would agree that this is simple. Very simple indeed. It is “die” or “divide”. Nonetheless the evolving features of these systems can be highly complex. One can try out different initial conditions, and see how the system proceeds. When the program is played, one notices clusters of cells (“populations”) and pulsating processes of near-extinction and sudden regeneration; one also notices how populations meet and reinforce one another. All this is beautiful in itself. But the most astonishing feature is the emergence of “gliders”, that is, localized structures that develop in one general direction and create exciting self-organizing structures that are far from simple. The Game of Life thus also models historical lines of descent, some of which continue to grow endlessly and continue to elicit new structures, new forms of order.
The question is now whether these computer-generated systems can be said to follow a few more general patterns. The seminal work of the physicist and computer scientist Stephen Wolfram has been devoted to this question since the early 1980’s (summarized in Wolfram’s A New Kind of Science , 2002). In order to be able to investigate the world of CAs systematically and unbiased, Wolfram chose the simplest possible CA, a one-dimensional CA with only two colors (black-white). Any step forward is then determined by only the three cells in the row immediately over the cell, which has to make a “decision”. The three upper row cells thus have only 2 x 2 x 2 = 8 possible combinations of color. Now with only two colors, the possible rules for deciding the next step for any cell are 2 8 = 256 possibilities. During his systematic search, Wolfram discovered the universal feature that all CAs fall into four main classes.
Class I consists of those CAs that simply die out very quickly. It is not difficult to predict that if the rules do not allow for enough reproduction of black cells, the screen will soon be all white, and vice versa. Formally expressed, the system fades away into a single “limit point” attractor.
Class II rules are a little more lively, but eventually they begin to oscillate repetitively between a few states. Even though we see no evolving logic, we can nonetheless discern distinctive nested structures, where smaller patterns are part of wider patterns. Formally expressed, they form a dynamical system as a two-point attractor.
Class III rules are more interesting in that they develop chaotic systems, though again with some self-similar structures appearing all-over, but in this case not repetitively. Class III systems thus display randomness, and look like some of the systems found in the mathematical chaos theory: The spontaneous evolution of CAs is neither derived from the initial conditions nor from a specific tuning between initial conditions and the mathematics of chaos . Rather, the random patterns are intrinsic to the class III rules.
Class IV , however, contains by far the most interesting features, which appear at the creative edges between the regular patterns of class II and the random patterns of class III. These are rare indeed, but quite significant, because they show that highly complex and â€˜interesting’ behavior can be produced against the background of very simple rules. Patterns here grow without coming to a fixed point attractor, without repeating the same structures, but also without displaying the randomness that characterize class III.
The amount of systems in each class, however, seems to correspond inversely with their interesting features. That is, around two-thirds of the 256 rules produce the infertile class I states, but around one-third of the patterns continue to grow, as we see it in class II, III, and IV. Only 14 % yield the more interesting patterns (Wolfram 2002, 57). However, in an evolutionary arms race, these were the ones to whom the future belonged!
Now the question is, Can CAs be used to model and understand religious evolution? Let me just mention two examples from the more recent literature. The British mathematician John Puddefoot (2002) has applied Wolfram’s Four Class typology to different forms of religious discourse. As he points out, the exclusivist claim of salvation within some religious traditions has the formal structure of a single point attractor of Class I: By contrast, religions seeking a sort of cognitive equilibrium with its environmental culture follow the oscillating patterns of Class II systems. More individualistic and eclectic forms of religiosity, such as New age, follow the pattern of chaotic systems, whereas the strongest candidate for a highly competitive religions may be found in Class IV, where we find that novelties in religious discourse emerge at the critical edge between Class II and Class III phenomena. Thus the recurrent pattern of internal (maybe even “doctrinal”) stability and continuous dissipation under the constant pressure of cultural inputs from other religions and culture may be seen as the strongest candidate for religious self-development.
My second example is the so-called “Jihad Model” (RenÃ© Thomsen, Peter Rickers, Thiemo Krink, and Christian Stenz 2002). This is a consistent attempt to use a cellular automaton model on religious evolution. The model is a so-called multi-agent system (MAS) based on individual agents. More specifically the artificial world consists of five general features:
(1) A world (represented as a sufficiently large, but finite two-dimensional lattice),
(2) 2 times 200 individual agents with the following four attributes:
(a) An individual location in space, by which each individual agent is surrounded by the eight neighbors in their immediate environments.
(b) An energy level between 100 as their upper limit and 20 as the lowest hunger limit, below which any agent has to prioritize the search for food.
(c) An age which is determined the remaining life-span, co-determined by the technological level of the culture of which the religion is a part.
(d) A religion with certain variable characteristics as defined below
3) Two (or more) religious populations are simulated in each experiment, and each religion is again characterized by four parameters,
(a) The enlightenment level influences (i) the maximum age of the agents (by 50 %), (ii) their combat strength and (iii) their likelihood of converting others to their religion.
(b) The aggressiveness level simulates the likelihood of combating a neighboring individual from another religion.
(c) The belief intensity determines the likelihood of converting other, or being converted to another religion.
(d) The birth rate is defined by the religious disposition either to mate and create offspring, or not to mate.
(4) The individual agents have the following five choices of action ,
(a) Mating (requiring two neighboring individuals of the same religion)
(b) Eating (consuming available food resources scattered around in the CA, thus upgrading the energy level of the individual agents)
(c) Attacking (thereby converting the other in case of superiority and downgrading the energy level of the former enemy)
(d) Converting (changing the religion of the other)
(e) Random walk (when no other rule applies)
(5) These actions, in term, are constrained by threshold values that represent the costs involved in the activities of mating, eating, converting, attacking and being injured. These threshold values are the variables that can be redefined from one computer experiment to another. One could say, for instance, that mating (and getting offspring) costs 50 energy units, eating 1, converting 5, injury 35 and attacking 2 energy units. On these assumptions, of course, attacking is modeled as a relatively risk-free strategy, which is hardly realistic on a battlefield, where the actors do not know the strength of the other part, and where wounds are not healed as fast as on the computer screen.
Much is debatable about the perhaps all too theoretic set-up of this “Jihad”-model (as it unfortunately is called). But still some unexpected insights came out the study. For example, it turns out that a religion with a belief intensity of 100, which at the same time also forbids reproduction (birth rate = 0), can still piggyback on the major control religion by way of continuous conversions. A pattern of population cycles emerges, very much like the pattern we see in the co-evolution of biological host-parasite relationships. In fact, the parasite religion here grows in periods of decline of the bigger religion, and vice versa. This looks very much like Class 2 CAs in Wolfram’s typology. – It would be interesting to see, if one could here further explore the model in order find examples of class III and IV.
Another interesting feature of the “Jihad model” is that relatively segregated geographic regions are continuously produced between two competing religions. Out of “individual” actions, clusters of religious communities are formed. The formation of ghettos, on this model, is a natural expectation.
I have discussed this model at some length here, because it shows that “hard” formalistic approaches are indeed applicable to more “soft” areas of study, such as religious evolution. Suffice it to say that there is a long route from observations to computer models and to the complexities of the real interconnected ystems. An improved model should therefore be able revise itself in the stress tests of being applied to real-world complexities. The process of computer modeling involves a long process of reality checks. Computational models will include a design cycle of observation, informal description, formal model, computer model, simulation, and least but not last: model verification by reiterated observation.
3. Complex Adaptive Systems (CAS)
What is missing from the CA approach is the function of learning that is characteristic for Complex Adaptive Systems (CAS), and no less for religious traditions. Many Simple Adaptive Systems have an internal program which controls the system-environment exchanges. Think of a thermostat, which directly adapts to the environment by controlling the input-output relations of temperature. A thermostat is certainly an example of organized adaptive complexity; however, it is not an example of self -organized complexity. The program of a thermostat does not develop itself under the influence of the environment. It connects directly, in a prefigured way, to the relevant aspects of its environment (“now too hot, now too cold”).
In the case of CAS, by contrast, a self-selective process takes place within the system. Inside the organism, an internal schema of the environment is carved out which is then — by trial and error processes — adjusted to the subsequent experiences of that system.
Wouldn’t this idea of complex adaptive systems be a confirmation of Neo-Darwinian selection processes? Yes and no. Yes, because a mechanism of selection is certainly at work in these quasi-cognitive processes; one could here argue (with Karl Popper and other proponents of evolutionary epistemology) that if an organism’s schema of reality is fundamentally misleading, it will soon begin to starve, have difficulties in finding a mate — and over time it will be outselected. But No, CAS also transcends standard Neo-Darwinian theory. For the interesting claim of CAS is that adaptation is something that happens at all levels of reality : at the level of the ecosystem (think of the emergence of the earth atmosphere of oxygen etc.), at population level (think of foxes surviving in cities), at the level of the individual organism (learning processes), at the cell level (think of the neurons in the human brain), and at the gene level (the prioritized unit of selection and reproduction in the received view of Neo-Darwinism). CAS thus transcends the standard biological view of adaptation, according to which adaptation is “a property of an individual organism, not of an ecosystem”, as John Maynard Smith has pointed out (in Pines 1994, 580).
Thus it seems that the idea of complexity may enlarge our standard picture of adaptation significantly. If learning processes take place at many levels (see also Weber and Depew eds., 2003), there are also many â€˜agents’ of evolution, for whom the enviromental influence â€˜makes a difference’. We are here approaching a biosemiotic view of evolution, according to which something (the environmental influences at large) means something specific (â€˜light’, â€˜food’, â€˜mating’) for somebody (an organism with internal, preferential schemas for orientation). Thus the idea of complex self-adaptation is structurally in accordance with the pragmatist Charles Sanders Peirce’s definition of a sign: A sign means something (reference) to somebody (the interpreter) in a certain respect (the context).
Now imagine that one were to regard specific religious systems as examples of CAS. One would then be able to identify certain internal programs that serve to stabilize the code of this or that religion, such as holy scriptures, recurrent liturgies, rituals (re-enacted at individual or communal level), and doctrines. In very strict religious communities, these programs will be used very much like a pre-set thermostat, that a priori determines what should be included, or excluded, among the environmental inputs. Imagine again, however, that the element of evolutionary learning or adaptation came into the focus of some specific religious community. In this case, the communal interpretation of the holy scriptures would come into focus alongside a reflection on the style , in which liturgies are performed, and on the use of rituals in given context. The words and concepts used in scripture, liturgy, and rituals may thus find different applications, including concepts of God and community.
Seen from a linguistic perspective, this would mean that the lexical terms (e.g. â€˜God’, or â€˜Buddha-Nature’) no longer possess a fixed semantic value, but are functioning like indexical pointers, the content of which will have to be specified within larger semantic discourse systems (“stories” or “myths”), the meaning of which again are co-determined by their use by specific groups or individuals. Semantic flows will begin to take precedence over against stable meanings. Since lexical terms no longer flow unsupported by discourse systems, and discourse systems gain their meaning from their particular contexts of usage, it is highly probable that scholars will be able to identify many examples of “mixed discourse”, that is, a discourse by which, say, Christian concepts are made fluid and understandable by concepts from other religions (often called “syncretism”), or elucidated by reference to secular sources of knowledge such as science (often referred to as “secularization”). In the light of the theory of CAS, such flows of religious insight would be more the rule than the exception; furthermore, the same patterns of evolution of religious concepts are likely to take place both in more “liberal” and in more “conservative” interpretations of faith.
Against this theoretical background, one could think of two distinctive types of research that may complement one another in the study of the cultural dynamics of religious evolution. One task is to show how the dynamic of religious evolution actually has taken place in historical communities, and how this dynamic is at work in present-day religious communities, where no religion is protected from external influences. One could here imagine important new studies of the history and sociology of past and living religions. Another research task will be to study, whether these actual processes of inter-religious and inter-cultural exchange will benefit the rationality and inner coherence of a religious tradition, or not. This type of research will demand a much more theological approach to the cultural evolution of religious traditions. It may well be the case that certain forms of rationality can be identified in the very process of passing on a religious tradition and communicating it to others. For in every communication, which involves human arbiters, there will be certain performance-based selection from the rich resources of religious tradition; some traits of traditional religion will be reinforced, whereas other traits will sink into oblivion.
However, both the concise historical or empirical analysis of the linguistic flows of religious discourse, and the philosophical or theological reflection thereupon will be able to learn from the formalistic approaches of computational complexity theory. Theoretical resources can be found within neighboring fields such as computational linguistics and cognitive science.
4. From Prisoner’s Dilemma to extended Game Theory (GAT)
One way to formalize such studies is game theory. Game theory is, like the theory of CA, based on individual agents, who (in a sort of contrived thought experiments) are imagined to perform specific strategies of choice vis-Ã -vis other actors. Game theory often shares the assumption of Rational Choice Theory, according to which actors make their choices by following the supposedly most beneficial (and hence “rational”) strategies for themselves.
Game theoretical analysis has often been formalized in the context of the Prisoner’s Dilemma , in which we have only two actors that are forced to share the same scarce resources within a closed setting, and have the choice between collaboration or competition. Since W.D. Hamilton’s foundational work on “The Genetical Background of Social Behavior” ( Journal of Theoretical Biology 1964), we have seen a suite of sociobiological studies aiming to explain collaborative behaviour as based on the sharing of genetical material (Hamilton 1964), or on reciprocal altruism without a genetic kinship (Trivers 1971, Axelrod 1990, 1997), or on an indirect reciprocity without a direct payoff for the individual (Alexander 1987; Nowak & Sigmund 1998).
What has come out of these well-known studies is a renewed emphasis on the importance of reiterated experiments so that one’s first choice may be altered under influence of the choices of the other agent(s). Thus if a collaborative behaviour is met by non-reciprocation from the other agent(s), it may be more advantageous to change strategy and defect the next time, or to expand the time horizon so that the other agent(s) are given new chances of collaboration. The shared assumption is that cooperation will benefit all actors, especially if one can find a strategy for stopping or punishing cheaters. Life and morality may be “non-zero sum games”, to put it with Robert Wright.
Reiterated Prisoner’s Dilemma’s thus give us a chance to model evolutionary learning processes on a relatively simple model. Even more interesting from a theological perspective, however, are the attempts to model collaborative behavior at the more complex level of social groups. One of the most influential and convincing studies in this respect are Eliott Sober’s and David Sloan Wilson’s Unto Others The Evolution and Psychology of Unselfish Behavior (Harvard University Press 1999). The convincingly show, first, that biological and economic research should separate the motivational issue of benevolence or malevolence at the psychological level from the functional issue of how to stabilize a social collaboration at the higher level of complex societies. Thus, they point out that social co-existence may stabilize the emergence of new moral codes by simply expelling or punishing individual cheaters within the systems. In so far as the keepers of moral systems (such as police and judges) are legitimized by the society at large, they incur, each individually, only modest personal risks in exercising justice; however, their job is quintessential for the functioning of the society as a whole. Second, Sober and Wilson have proposed simple mathematical models that show, why one must transcend the realm of genes and individuals in order to understand the cultural dynamics of human societies. At the same time, empirical psychologists have shown, that human persons, as a matter of fact, disgust cheaters, and rather want to sacrifice own benefits than allowing social cheaters to win their game. Both at group level as well as at the psychological level, there seem to be inclinations towards doing the good rather than just that which is of direct or indirect benefit to oneself.
As is evident so far, sociobiology and evolutionary psychology has been concerned about explaining moral behavior, especially the possibility of altruistic and generous behavior (Stephen Post & colleagues; NÃ¸rretranders 2002). Why not extend this research program into the field of religious evolution, including the notion of God and Ultimate Reality? First steps have already been done. David Sloan Wilson has applied his method on the issue of religious evolution, using the development of Calvinism as his historical test-case (Wilson 2002), and also many empirical studies of the psychology and spirituality of forgiveness have been presented (e.g., Worthington 2002). However, both the biological and the economic communities are divided on the issue, as to whether the individualistic perspective is sufficient to explain social behavior, or one would need to understand cultural and religious evolution at the more complex level of group behavior and religious semantics. The field of “Competitive Dynamics and Cultural Evolution of Religions and God Concepts” is an invitation to take part in this scholarly debate, if possible at more complex level than has been reached so far.
5. The Theory of Autopoietic Systems (APS)
Allow me to end with a note on the perspectives coming from autopoietic systems theory , or the theory of self-productive systems, which seems to me especially applicable for reconsidering religious notions of God, and of the human participation in divine creativity.
The general idea of self-organizing systems is sometimes prematurely equated with the notion of autopoietic systems. The difference is that while self-organizing systems combine great variability with internally regulated mechanisms or programs (as we see in CAS), autopoietic systems produce new internal components and thus continuously create new system environment-interactions. While the concept of self-organization still retains the idea that systems are organized out of pre-established elements, the concept of autopoiesis more radically contends that the components themselves may be created only inside organized super-structures. Self-transformation extends not only to the organization of the system but also to the elements specific for that system. It is only in a cell, for instance, that we meet the special arrangements of molecules that make up its membrane. Or, again, consider, the how the carvings of the brain (like physically engraved schemata) are produced in a kind of “topobiological competion” (Edelman 1992, 83), that recurrently reshapes the neurons and their interacting networks. Selection processes thus take place also in the brain, to the benefit of the brain’s over-all plasticity.
In autopoietic systems, therefore, there is no separation between producer and produced. A cell’s being is given only by virtue of its internal dynamical operations and the system is not a substance definable prior to its operation (immune systems therefore vary significantly in genetically identical twins). It is the internal functioning of the system that both determines whether or not the cell should build up new elements, and how the cell picks up (or ignores) specific elements of the external world (Maturana/Varela 1992 (1987), 43-52).
Taking the feature of complex adaptability seriously means taking seriously the pluralistic order-and-disorder of nature. The world has many centers of control, and to each is assigned a certain process autonomy. Like other types of complexity theory, the theory of autopoietic systems presupposes a constitutive materialism (“there exist no other elementary particles than those known by the physical sciences — or in principle knowable by them”). However, what are important are not the singular objects (e.g. atoms or molecules), but the work cycles they perform within holistic, yet highly specialized networks. What matters is not the generic amount of matter’s physical energy, but the specific physical organization of matter.
The pluralistic order-and-disorder has its ontological basis in the operational closure of the different systems themselves. That is, a system is not acting at the mercy of the environment, but is itself determining what is relevant, and what is not relevant in the surroundings. Accordingly, there does not exist one objective environment, common to all systems, but there exist as many environments as you have adaptive systems. Autopoietic systems may react to their environments on all grades from negative feedback (balancing each other) to pos tive feedback (mutual