Complexity Study Group, 22 February 2000

 

‘Biological Evolution as a model For Social Systems’

22 February 2000

Abstract: What can the rise and fall in species origination and extinction, the fluctuation in market prices, or even the pattern of the human heartbeat tell us about the dynamics of complex systems? Mathematical analysis shows a distribution, where the frequency of event relates to size and follows what is called a ‘power’ or ‘scaling’ law. But such insights tell us little about what is going on to cause such events. Deterministic chaos theory tells us how random events can occur in physical systems but little about self organisation and adaptivity in organic complex systems. Complexity

science often fastens on biological evolution models but struggles to explain the role of rational agents in human social organisations where knowledge transference becomes a major determining factor. The discussion led by Brian Goodwin covered aspects of complex systems, ‘memes’ as units of cultural transmission, natural kinds and the move towards a science of qualities. Contributions included searches for a better understanding of the terms ‘evolution’, ‘co-evolution, ‘closure’, ‘emergence’ and in human organisations, ‘knowledge’ transfer. There were also reports from ongoing research into aspects of social and business systems .

Director of Research : Eve Mitleton- Kelly

London School of Economics

Houghton Street

London WC2A 2A

Coordinator : Prof. Brian Goodwin

Schumacher College

Devon.

Compiled For The L.S.E.

by Geoffrey J.C. Higgs 22/2/00

Introduction

Once, scientists tended to think that with sufficient information, all natural systems were predictable and controllable. This ideal was eroded, first by fluid dynamics and later by Lorentz’s study of weather systems. Even if mathematical expressions could be found to model the dynamics, sensitivity to initial conditions meant that predictability often rapidly diminished as the amount of data required for prediction went up. This was not to say that stable patterns do not arise in such complex systems but the term ‘emergent’ has come to signify properties which cannot be predicted from what goes on in different parts of the system. Thus the term is a convention indicating the above but not excluding the possibility that subsequent ‘reduction’ might be possible. Many systems, however, show self organising properties and point to a kind of creativity that cannot be explained in a deterministic way. It is for this reason that many researchers have resorted to evolutionary models often taking Darwinism as the starting point. But Darwinism or even neo-Darwinism is not always applicable to organic systems and attempts to use such models as analogies for human social organisations are the subject of much debate.

Features of Complex Systems

Palaeontologists studying the fossil trace left by extinct species found a random pattern of extinctions over time and whether the data was coarse in terms of families or fine in terms of species a similar pattern was found. Such self similarity is a feature of a number of mathematical structures such as Lorentz’s ‘strange attractor’. Rickard et al found that the distribution of extinction size plotted against frequency followed a ‘power law’ such that the relationship was Y = 1/Xn where n is some power (in this case close to one). Moreover a log/log plot gave approximately a straight line indicating that as the size of the extinction increases so it becomes rarer.

These observations provided a means of looking at other systems where ‘power laws’ or ‘scaling’ might apply. The food web in the North Sea around Nova Scotia is a typical complex ecosystem. There is a very rich pattern of interaction between species. Some of these interactions are positive in that a species derives benefit in the presence of others and some negative in that it doesn’t. Darwin's original idea of ‘survival of the fittest’ was probably influenced by nineteenth century Whig philosophy and the idea of the ‘free market’ and painted nature as ‘red in tooth and claw’. But though he was right about competition this seems only a partial description of the evolutionary process. In a trophic web there are primary producers which do not necessarily experience wholly negative interactions and in fact develop many positive interactions between themselves. In nature it is often not a case of doing better but doing well enough. Such an aim is now prevalent in successful business organisations. Unbridled competition leads to mono-culture. Healthy stable systems need variety.

What Rickard and colleagues did was to model the evolutionary process by assuming a complex network of interactions between agents as individuals which was set up as a random probability choice. For example, a number chosen randomly between +1 and -1 would represent the degree of interaction between two species. Once this has been done between all species in the system the process is allowed to proceed and at every time step other connections and degrees of interaction are chosen. This is very similar to Stuart Kauffman’s random Boolean networks which have positive and negative interactions and ‘amplification’ of positive or negative effect. In Rickard’s model the agents are purely defined by their relationships and have no other interesting properties. A species is deemed to be extinct when the net sum of the interactions is below zero. Each time step constitutes a new field for each species and when extinctions occur the ‘empty niches’ are filled by existing species at random rather than on a ‘best competitor’ basis. Models which try to simulate what is seen in the fossil record don’t seem to work if you choose the species with the highest fitness and though random choosing is an approximation it allows for many factors that have been left out. The model works on a coarse level and assumes some sort of fitness and phenotypic effect at every time interval. It’s possible that extinction could be zero, up to the total food web. Human social systems are of course different. If we look at the price changes on the money market it looks like a random process though we could assume that people only buy stocks and shares with good reason (even if it is because the price is falling).

‘Power log scaling’ and ‘self similarity’ seems to be a deep generic features and we don’t really know what it means in terms of the dynamics. If we look at the size of cities these seem to show a scaling distribution : the number of large cities is small, the number of small cities is large and the distribution follows a straight line on the log/log graph. Although we consider ourselves rational agents we are overall just a species interacting with the environment in the way other species do and because we are not at present building cities according to any pre-defined plan a city grows ‘organically’. In fact images of population density look very organic; like ganglion or nerve structures. It also doesn’t seem to matter what range of sizes is considered. They all show the same kind of distribution.

Per Bak’s ‘sand pile’ is a model of a physical system which displays ‘criticality’ and fluctuations in behaviour which mirror the extinction patterns. If sand is dropped to form a heap at some point in time the slope of the sides becomes such that a single grain can produce either no avalanche or one that may run the whole length of the pile. The distribution of avalanches matches that of the biological extinctions. It’s a kind of ‘self organising’ criticality that seems to keep coming up but there is no mathematical characterisation of the intrinsic properties. The gradient forms a stable attractor and events vary away from it in the manner described.

Species in biological systems tend to converge after a number of fluctuations onto a mean (population)value. Big fluctuations occur after an extinction event because of dramatic changes in interaction relationships. But large extinction events do not necessarily require long periods of recovery. Prigogine looked for features which characterised a ‘far from equilibrium’ domain in terms of energy and matter flowing through it and this may still provide the best way of understanding such systems. We can characterise some systems in terms of interacting ‘loops’ or ‘cycles’. Strange attractors for example, can be characterised in terms of divergent and convergent flows.

Avalanches on a sand-pile may not seem a very interesting example of random but patterned variation but similar fluctuations occur in stock market prices where people are assumed to be rational agents. There is again no characteristic time scale over which events are likely to occur. People have in the past identified periodicities but these have never proved reliable. However if we look at past events such as the 1987 crash then retrospectively we can see that there was a trend and a ‘quasi equilibrium’ in the market. Divergencies are governed by some kind of law though we do not know whether this is some kind of deterministic chaos or otherwise. But it has some degree of order and we can make reasonable predictions over short periods of time in much the same way as forecasting the weather. Yet we do apply epithets to systems that are indicative of the behavior of the people in them such as a ‘healthy’ or ‘fevered’ markets.

What are the optimum conditions for an economy? Intuitively it seems that the small local economies are stable if a bit static, whereas a global situation forces innovation at a pace which causes instabilities. The question is ‘How do we achieve a balance between stability and innovation? There has been much debate in complex systems theory about whether diversity of interactions gives rise to stability or instability. The original intuition was that the richer the connections the more stable the system but Bob May did a linear analysis which suggested the opposite. This shifted opinion and it has subsequently been found in detailed ecological studies that what is required is both diversity and a rich interaction. Global economies that are driven by a few large international companies are not very stable.

Genes and Memes

Whatever objective characterisation of complex systems scientists may come up with, we as human beings live within complex systems and co-evolve with others. In human organisations it is relationships that are important and the transference of knowledge within the cultural system. In a further application of biological evolutionary theory, Richard Dawkins introduced the concept of ‘memes’ as units of culture transference in the last chapter of his book ‘The Selfish Gene’. Memes are the cultural equivalent of genes as the units of hereditary transfer and may take the form of ‘tunes,ideas,catch-phrases, clothes fashions,ways of making pots of building arches’. Such units are transmitted from one person to another like a virus which replicates (and modifies) as it goes along. It is questionable how far this analogy can be pursued. As Dennett says in ‘Darwin’s Dangerous Idea’: ‘There is no denying that there is cultural evolution in the Darwin-neutral sense that cultures change over time, accumulating and losing features while also maintaining features from earlier ages.... But whether such evolution is weakly or strongly analogous to, or parallel to, genetic evolution,.... is an open question’ (p.345).

Genes cannot simply be identified as particular hunks of DNA and are at best a vast oversimplification. But as Dennett says on page 59: ‘..no one wants to throw away such a valuable tool’ and ‘We may, if we like, talk of virtual genes, considering them to have their reality distributed around in the concrete materials of the DNA’. Genes are not the sufficient cause of all the properties of organisms but do account for heritable differences. Brian Goodwin does not regard the ‘meme’ concept as particularly useful since it is not accessible to scientific investigation in the way that genes are(even if they are ‘virtual’?).

There is an article by Mary Midgely in the ‘Journal of Consciousness’ entitled ‘Being Scientific About Ourselves’ where she examines the possibility of using memes to explain cultural history. She takes as an example the ‘witch craze’ in Europe for which the meme explanation is to say that the meme came along to infect a population that was not properly immunised against it. It lasted for a time and then disappeared and that nobody was responsible for it. Brian Goodwin regards this as a preposterous view of history that explains nothing. It tells us nothing about what else was going on in European history at that time; the source of fear, the status of women, for example. Nevertheless memes do seem to have some independent existence in cultural transmissions. Dennett himself refers to an embarrassing occasion when he found himself humming the tune ‘It Takes Two To Tango’ which he describes as a dismal bit of chewing gum for the ears. (Max Boisot thinks the meme may be useful as a vehicle of information transmission- see ‘contributions’)

However just as a gene has to be interpreted, so does a meme; it ‘beds down’ differently in different contexts and a particulate explanation of the cultural transmission process leaves out the hermeneutic or interpretative aspect. It also fails to connect with the concept of agency. The reductionist trend of traditional science assumed that it was possible to understand the ‘whole’ by considering the ‘parts’. This may be a good way of explaining differences between things but not the uniqueness of a thing itself. Even the word ‘organism’ dropped out of modern biology perhaps because it smacked of ‘vitalism’. The concept of ‘health’ has tended to drop out of medical practice though it has hung on in holistic medicine. Such trends result from an overemphasis of the reductionist paradigm. The belief that the meme is somehow context free is probably due to the success of the gene theory and is a warning against an over-zealous attempt to overlap biology onto culture. We have to connect the meme to context to understand emergence in social organisations. We have to understand ‘qualities’ and values in order to understand why agents do the things they do. Brian Goodwin thinks that a science of qualities can perhaps be gained by a method of inter-subjective consensus which can link knowledge with ethics or aesthetics. If you can’t get measurement or assessment in some form you don’t get qualities or values and you don’t get explanations for them.

Towards a Science of Qualities

The science of qualities can perhaps be introduced via the concept of ‘health’. Complexity theory has already suggested a different way of looking at health in studies of heart rhythms. The electrocardiogram registers the electrical impulses at the surface of the body which result from heart contractions. In the late 1980’s when complexity theory was just beginning to be put to good use in scientific technology a cardiologist working in the Bethesda Hospital in Boston suggested that heart condition might be better understood by the application of chaos theory. The condition of cardiac arrest is due to high frequency contractions in which the blood is not pumped. Though heart rhythms are not classical examples of chaos theory the suggestion prompted people to re-examine the data in terms of mean values of heart rate and the differences in the patterns which enable pathological conditions to be recognised. By looking at the variations in the inter-beat interval it was possible to tell the difference between a healthy person and one suffering from a pathological condition. Significantly it is the healthy person who shows the random pattern and the person at risk the one showing too much order. The conclusion that too much order was a sign of danger was startling in a medical tradition that looked for ‘homoeostasis’ which is the tendency towards a relatively stable equilibrium between interdependent elements. Though there is variation around a mean such as in the circadian or diurnal rhythms of body temperature or in the pH of the blood or the concentration of ions this was something different. It also turned out that the signature displayed self similarity so that a record on a coarse time scale when refined to that on a shorter time scale showed similar random patterns. The technique is now beginning to be used for other conditions such as sleep apnoea and what is revealed is that all healthy people share the same scaling law with respect to the inter-beat interval fluctuations.

The conclusion that the random pattern of heart rhythms is a sign of health is an important one. It emphasises that the chaotic signature is a necessary part of nature and not an aberrance and it points to complexity science technology. The heart exists in an unpredictable environment in which its owner may make sudden movements or experience strong emotion. It therefore dances to a chaotic tune; always ready to move and alter its pace. The signature is complex because although it is chaotic it is not the sum of utterly random influences and is therefore an emergent property of the whole. There is a kind of self organised criticality in much the same way that an ecological web is an emergent property of the interacting species.

Such an indicator is both a quantitative and partially qualitative indicator of health. You could have some measure of how much a condition differs from some kind of norm but you could also intuitively assess its significance in much the same way as say eye brightness or hair condition. When the average general practitioner diagnoses a sick individual he or she should take account of quantitative data such as pulse rate, blood sugar, or different blood cell count but these are not sufficient indicators of health. An efficient doctor should pay attention to say complexion, or tone of voice, posture and personal history. Chinese acupuncture diagnoses via a system of pulses. For each position there are three types of pressure used to feel the pulse which is indicative of the state of the entire body in nine different ways or variables. Again, this is a semi-quantitative semi-qualitative method. Naturalists use bird-song as an indicator of the quality of habitat. This is intuitive or tacit knowledge. It is only because we are so used to the analytical method using only quantitative data that we ignore the possibility of using qualitative information. We are so seduced by traditional science that we believe that it is only quantitative data that gives us reliable information about the world.

Qualitative observations are assumed to be unreliable because they are supposed to be purely subjective and intuitive but individuals can reliably come to conclusions about the quality of experience of other beings including other species. In our dealings with other people we rely on qualitative evaluations and a ‘theory of mind’ all the time. Traditional scientific method based on a strategic move by Galileo and a theory of atoms has been extraordinarily successful but a study of complex systems shows that we may be reaching the limit of its explanatory usefulness. Doctors assess their patients, we assess other people and we must read the person quantitatively as well as qualitatively in order to know how we must behave towards them. We are participating in a very active way in the condition that is presented. An good ecologist participates in a very active way in the ecosystem he or she is assessing. A good farmer participates in a very active way in the quality of life that his or her animals are experiencing. But none of this kind of intuitive knowledge is regarded as reliable in traditional science.

A study carried out on the welfare of farmed animals by Francoise Wemesfelder for the Scottish Agricultural Office in Edinburgh suggests that we can gain a reliable knowledge of the quality of life of pigs by studying them but not in a behaviourist sort of way. A behaviourist would count the number of times a pig snorted, how many times it patrolled its pen or went for its food. But this is only part of the way we relate to pigs or for that matter, to each other. Francoise had a group of pigs that had all grown up together. Each of these was then introduced separately to a standard pen and a researcher with whom they could interact. Each researcher was then invited to record descriptors which they felt were appropriate to the pig, such as ‘boisterous’, ‘pushy’, ‘laid back’, ‘aggressive’ or ‘cool’; anything that they felt was an indication of the quality of life that the pig was experiencing. The project used 18 people working independently this way with 20 pigs. All the terms for all the pigs were then written out with a scale of 0 to 100% drawn beside them. The researchers were asked to indicate the degree to which the animal manifested the term. If a particular term was not used of a pig then the indication was counted as zero. This method is called ‘free choice’ profiling and is used for assessing the quality of food or drink for humans. The data is then processed onto a matrix constructed according to the number of terms used by the researcher who used the most terms. Statistical analysis using mathematical rules of reflection, inversion and averaging is then used to find clustering. A profile is then built up which shows a remarkable degree of consensus. It is of course necessary to assess whether people agree the same definitions of the terms used but again an assessment of semantic clustering can be found using the same methods. The conclusion is that this represents a way in which we can begin to explore our ability to read quality. If we are going to have a science of qualities then we do need inter-subjective consensus.

Consensus is not just an assessment of the collective opinion at any one time. Any person is at liberty to disagree and go through a process of convincing other people of more appropriate terms of assessment and this progresses the science by shifting the emphasis from one focus to another. This method is universal; if we get 20 different therapists to come up with a diagnosis of an individual the diagnostic powers of either the individual or the collective could be measured against some previously established standard. It may be that, as far as assessing pigs is concerned, the anthropomorphic projection may present a problem and some people are more experienced than others. In medicine it is acknowledged that there are good and bad diagnosticians and the way a person learns to diagnose is through experience and the systematic cultivation of the intuition. People say that medicine is as much an art as it is a science though we should strive to make it as ‘objective’ as we can. In science we use intuition in making sense of data; which is to say we put quantitative data together to construct a coherent whole. But scientists are currently trained into western scientific discipline which is a science of quantities. What used to be called natural philosophy is perhaps closer to what we have been talking about, though there are a number of reasons why the consensual approach may prove difficult. First there is the difference between the language of ‘participants’ and the language of ‘observers’. On a local scale it may simply be that researchers interpret data in a different way to the outside community (because of anthropomorphism or the degree of artificiality of the situation for example) but on a larger scale this is really an ethnographic problem.

One of the reasons for adopting a quantitative approach is to simplify communication. Although we do not have a clear idea of how we formulate theory in science it has been described as proceeding via the ‘hypothetico-deductive’ method. This involves postulating some hypothesis about what are thought to be the critical attributes of some system and then examining the consequences by observation or experimentation. Max Boisot thinks that what is perhaps limiting with this method is that if our general aim is to establish some basis for inter-subjective objectivity and be able to communicate with outsiders then maybe traditional science does it prematurely (or from a narrow standpoint?- Ed). We are starting from something qualitative and then structuring it so that it can be codified. A more ‘inductivist’ approach would be to home in on the critical attributes from a less rigid standpoint though we still need to structure and codify so that we can communicate. We need to standardise and we need to ‘objectify’ so that communication doesn’t go away, but the basic problem is that in the process we lose much of the richness of the qualitative data which from some different standpoint may be relevant. The traditional science approach is however not just about communication but control. What dealing with complex systems teaches us is that ‘ wholes’ have their own natures.

Natures and Natural Kinds

We understand that pigs, for example, have a certain nature and ‘natures’ don’t exist in contemporary biology. In Holistic science there would be such a thing as ‘pig nature’ which is an emergent property of the pig system. Moreover if you understand ‘pig’ to have a certain nature then you will behave towards it in a certain way just as if you understand that it is the nature of children to play then you will provide conditions for them to do so. In saying that ‘pig’ is a ‘natural kind’ we have moved beyond the bounds of contemporary science to philosophy. Of course ‘pig nature’ is highly flexible and not fixed. Individual pigs have their own version of that nature and there is also the problem that once you put the pig into a certain context, e.g. with its researcher where it develops a strong relationship, then the study is biased. There are a number of epistemological as well as ontological issues here. For example, how do we balance consensuality against expertise? There is quite a lot of doubt in society about so called ‘experts’ who may not have a very good appreciation of, say, the quality of life of pigs.

One of the ethical considerations of biotechnology is the potential ability of people to alter natural kinds. Darwinism has nothing to say about natural kinds. Neo-Darwinism reads species in terms of genes such that each species is defined by a set of functional characters that allow it to survive and any set of functional characters is possible in subsequent evolution providing that condition is fulfilled. There are no boundaries for species so any organism can theoretically be changed in any way. This is why contemporary geneticists assert that there is no difference between biotechnology and evolution. Any variety that human beings can dream up, providing it can be perpetuated, is OK. And if turkeys have so much breast meat they cannot stand up or reproduce, well that's OK, we can do it for them. But organisms are complex adaptive systems that we interact with and co-evolve with and as such deserve some respect. In reading organisms as natural kinds and as wholes we can still allow an immense diversity of form or behaviour. But if we go outside that domain, which admittedly we cannot define then we may transgress. If using our intuition we can say that a pigs quality of life is bad then then we have a duty to alleviate its suffering. But ‘suffering’ is not a scientific term, nor is ‘pain’ though we need such terms to help us draw the boundaries. We know, just as we do for humans, that if we deprive pigs of certain forms of behaviour we are actually violating their natures. If we manipulate genomes to the point where the species is clearly unhealthy then individuals suffer. The public tends to feel this is not legitimate and maybe we should trust the public and put intuition before scientific ‘experts’.

Bill Torbet has an article called: ‘Towards a Science of Qualities in Organisations- Lessons from Complexity Theory’, in the ‘Journal of Concepts and Transformations’. He suggests that ‘knowing’ starts with a relationship between oneself and others. It is not just an academic skill but one of amity and creative meaning. Such knowing requires an intuitive knowledge of purposes, an intellectual knowledge of abstraction, an embodied sense of ones behaviour and an empirical knowledge of (the other side of the world?). Complexity theory that does not take account of human agency, qualities and values cannot give rise to a science of organisations. We have to shift from a paradigm that seeks to understand in order to control, to one of qualification.

It was suggested that fuzzy logic’ which is an improvement on the logic of categories might provide a means of escaping from a strict hyperthetico deductive or rigidly codifying approach. But though this is a much more fluid and realistic way of organising logical categories, qualities such as the experience of ‘red’ or ‘blue’ are much more immediate. You don’t use a collective consensual response to decide whether your partner loves you. Scientific culture does need a crutch to go from existing methodology to an appreciation that an evaluation of qualities can prove reliable and we may even dispense with the desire to put everything in terms of mathematics. If ecologists reading the landscape say ‘don’t put a pig farm here’ then we will learn to trust them.

In evaluating the evolution of Homo Sapiens it is important to see the biological and cultural aspects as a whole. There has been a great tendency to consider that there were several stages of ancestry and then suddenly an anatomically modern human emerged in the form of Homo Erectus and then there was Homo Sapiens who was morphologically fixed. Biologists tend to assume that modern man came into existence at such and such a time and from then on it was just cultural evolution. We have to be careful in talking about separating biological and cultural evolution and remember that Homo Habilis would have had the limbs and balancing organs to ride a bike if he’d had one. You can’t even distinguish biological evolution from geophysical evolution. The movement of tectonic plates may depend on limestone as a lubricant which is produced by life and yet we think it’s just physics when we think of plates floating on a liquid magma core. The Gaia Hypothesis intermingles biology and geophysics in a way that cannot be separated and this indicates that the boundaries are breaking down. Rate of cultural evolution may be greater than physical evolution but our physical abilities depend upon our cultural and this is what Holistic science is all about.

Contributions

(As with all discussions, the farther people are away from a microphone the more difficult it is to hear what they are saying but I have tried to capture the essence of people’s contributions so that we can start off next time on a good basis. I also apologise here and now if I have misattributed anyone and if I have seriously misrepresented someone then I will either put out an amendment or the person can point out my errors with a presentation next session- Ed.)

In ordinary discourse the term ‘evolution’ is often taken as merely a synonym for change. Eve Mittleton Kelly wondered whether we could define the terms ‘evolution’ and ‘co-evolution’ more closely in a social systems context. According to Pianka(1994) ‘evolution’ in biology refers to temporal changes of any kind and ‘natural selection’ is only one way in which it can be brought about. Natural selection is the important force in evolutionary change because it results in a conformity between organisms and their environments, but other possible mechanisms include the inheritance of acquired characteristics, gene flow, meiotic drive and genetic drift. The Shorter Oxford English Dictionary defines ‘evolution’ as ‘development or growth as of a living organism’ and also stresses ‘growing’ as opposed to ‘being made’ and that the process is an ‘unrolling’ or opening out’.

In a human organisations the changes seem to include major ones imposed by specific individuals as well as all the micro changes that are occurring all the time through the interaction of all the individuals who make up the whole. Mc Kelvey’s study: ‘Visionary Leadership vs Distributed Intelligence: Strategy, Micro-evolution, Complexity’ (1999) deals with the micro level of interaction between individuals and between individuals and artifacts (I.T.) and also the macro level where there is an interaction between the business and the information net or environment.

The term ‘co-evolution’ as defined by Ehrlich & Raven 1964, Pianka 1994, Kauffman 1993 & 1995, McKelvey 1999 a & b and Koza and Lewin 1998 is that the evolution of one domain is partially dependent on the evolution of the other or that one domain changes in the context of the other. The idea is that co-evolution depends on an evolution of interactions or reciprocal evolution (Futuyma ,1979). One hypothesis being studied at the LSE is that the intensity and density of interaction between two entities affect the rate of co-evolution between the two domains.

Pierpoalo Andriani of Durham University is interested in ‘closure’ in a complex evolving system and also whether the acceleration of a co-dependent evolutionary process is proportional to the degree of variety in the micro environment and the intensity of interaction. ‘Closure’ occurs when there is a de-coupling of some part of the evolutionary process such that it becomes self referential. Jacques Monod, for example, suggests that there is a relationship between the evolution of the human cortex and the development of language. Increasingly complex behaviour patterns necessary for the individuals of a hunting party, to coordinate their efforts, required an increasingly sophisticated language which in turn required an increase in brain size. The remarkably fast development of cortex and language constituted a ‘positive feedback’ in the evolving biosystem. ‘Closure’ in this sense can occur wherever there is a higher transactional activity in some part of the evolving system and de-coupling leads to the emergence of new ‘wholes’ with emergent properties. Closure is probably therefore a precondition of co-evolution.

In the socioeconomic environment increasing interaction between companies occurs when:

1. When the organisation ceases to be vertically integrated and there is flexible specialisation.

2. Where there is a high density of user-producer relationships.

3. Where there is a high density of products and processes and complementarity between organisations.

When ‘loops’ of interaction occur in a system the non linear behaviour gives rise to a full repertoire of complexity effects such as ‘bifurcation’, ‘increasing returns’, ‘recycling’ and ‘multiplier effects’. The multiplier effect is dependent on the degree of variety and the extent of internalisation has been modeled using the equation shown in the diagram.

 

Image: wpe7.jpg (23906 bytes)

 

Total economic income (Y) is plotted against income from external transactions (exports)(X). The figure shows that bifurcation occurs at a certain value for X.

Duncan Bell intends to use the idea of a ‘complex adaptive system’ to explain the evolution of world politics. In Murray Gell-Mann's approach (1995), a complex adaptive system is one that finds regularities in a stream of incoming information and compresses the regularities into a schema (model) which is then used to describe/understand the world and prescribe the behaviour of the complex adaptive system itself. Depending on its success in dealing with the world a CAS either perishes or survives. The approach avoids determinism and teleology and explains how systems cope with chance and variation in the environment. It can also lead on to the important idea of agency and intention. Immanuel Wallestein in ‘The Heritage of Sociology, the Promise of Social Science (1998) uses a world systems analysis that does not separate the economy, culture and politic in what is called a unified ontological approach. Because most world systems analysis lacks a suitable generative mechanism Duncan proposes to use complexity/evolutionary theory to explain the development of social and political structures throughout history.

Ted Fuller and Paul Moran explore the application of the ‘evolution’ concept to small and medium sized business enterprises. They define the term as a construct grounded in inter-generational differences. This is close to neo-Darwinism and sees the process as one generation evolving from a previous one through some form of genetic cross-over or occasional mutation resulting in a bifurcation of species as natural selection weeds out those organisms less able to meet the demands of the environment. The process is not linear. The ‘Cambrian Explosion’, for example, marked a peak in the rate of bifurcation of species (Gould 1989). The implication is that far from ever existing in a steady state, life organises itself spontaneously into a characteristic and much more precarious one which does not require external cause to explain its emergent features (cf. Bak-1996).

The adaptation of complex systems to their environment is a form of learning which implies a more ‘conscious’ or sentient response based on memory. Whatever this mechanism might be (in reality) Kauffman understood it as resulting in a fitness landscape: a descriptive model of the relative fitness of the different inter-connected actors in a system (Stuart Kauffman-1995).

‘Evolution’, ‘adaptation’ and ‘emergence’ are notions in dynamics and share the common characteristic that they are irreversible in time. Unlike say, a linear Newtonian pendulum system, the complex adaptive system does not move back to an historic position or state. Any position is the result of a past history which creates the conditions that project the CAS on a particular trajectory and gives what Allen (1997) calls an ‘evolutionary tree of successive structures’.

The central idea of complexity is that interactions between the parts of a system create novel, unpredictable patterns and whilst a system’s history is relevant in understanding its dynamic, isolation of a system’s parts does not reveal the causal mechanisms. Representation by modeling distinguishes objects in the system and describes the relationships between objects. The non static nature of these relationships and the notion of adaptation implies a motive force which leads to the concept of agency (Axelrod (1997); Casti (1997). Individual agents therefore, make up the population of a system and are in the receipt of local information which is different from that of other individual agents since no agent has an overview of the entire space. Each agent is ‘intelligent’ in that it uses ‘rules’ of behaviour in adapting and is able to change the ‘rules’ by which it acts. Such a description of interacting agents enables modeling via inter-generational relationships and this is resonant with current social theories of structure and agency (Giddy-1984)

It may be possible to see the entrepreneur or the small business as an adaptive agent in this way though a small business or enterprise may not be an appropriate ‘unit of analysis’ because of its multi-layered structure. Gillies et al (1998), for example, used the notion of the unit of expertise as a unit of analysis in industrial co-operation between small enterprises. Greenfield and Strickon (1986) and Low and MacMillan (1998) draw an analogy between population ecologies and entrepreneurship and evolutionary and ecological metaphors of emergence, fitness and mimicry resonate with observations on small enterprises in the economy. Such enterprises vary enormously in ownership, location, size and sector activity but each has its own ‘initial conditions’ and incurs a number of ‘accidents’ in its temporal path. Businesses operate their own ‘rules’ as well as complying (more or less) with more general rules. Business strategies can be seen as aiming towards some form of ‘niche specialisation’.

In acting as part of a bigger economic and social system, businesses operate in an environment regulated by key economic stake-holders such as banks and government agencies and these provide at least some of the rules of behaviour. Swarming is commonplace both physically in terms of location (Gillies et al (1998)) and in the use of particular technologies (North, Leigh et al (1991)). Energy in the form of cash, flows through the system, with those enterprises not maintaining cash flow ceasing to operate.

Fuller and Moran suggest that the notion of agency ( implying adaptation, evolution, fitness, interdependence) coupled with a theory of evolutionary autopoietic structures provide a frame of reference for studying enterprise dynamics such that:

• Adaptation is an over-arching concept of learning, memory and change over time.
• Evolution is an over-arching concept of alternatives, bifurcation, diversity and selectivity.
• Fitness is an over-arching concept of goals and relativism.
• Inter-dependence is an over-arching concept of relationships and co-evolution between agents in the system.
• Each concept is underpinned by temporal relativism; the arrow of time.

Longitudinal empirical observations of systems can give descriptive accounts of dynamics and ‘actors’ strategies, together with descriptions of competitive ‘fitness’ of an organisation or sector or entrepreneurial adaptation. The ontological adequacy of such a description is often assessed by its instrumental reliability; if for example it helps explain a sensed phenomenon or accords with ‘actors’ perceptions. However identification of causal mechanisms in systems (Bhaskar-1978) and differentiating these from accidental regularities is weak. Identifying salient features in systems is necessary for theory building but insights may be theoretically ‘accidental’ even if instrumentally adequate. Fuller and Moran ask whether complexity theory could provide ways to examine the theoretical adequacy of such insights.

Nigel Phillips suggests that ‘division of labour’ in social groups is a natural evolutionary development that provides a means of privileging offspring so as to increase their chances of growing to maturity and producing offspring of their own. Though the potential for such behaviour must exist within the genetic makeup of a species it is evolved and reproduced through the social dynamics of the group. Structural coupling between members creates an operational closure that maintains the group as a viable entity and knowledge passed to junior members perpetuates the structure. The process by which working knowledge is acquired has been described as ‘Communities of Practice’.

The dynamics of ‘communities of practice’ is quite different to the division of labour that dominated 20th Century organisations. In the past approaches to job design via ‘scientific management’ sought to remove the need for workers to adapt and for jobs to evolve by dictating ‘optimal’ behaviours for every situation. It was no part of a worker’s job to adapt to change or reproduce knowledge (of this change) to subsequent generations of workers. The failure of such a mechanistic approach to job design in organisations and the inability to cope in the face of rapid technological change has led to calls for flexible, adaptable workers capable of continuous lifelong learning. The conceptual framework of autopoietic closure and communities of practice offer useful insights into how the image of an adaptable, empowered knowledge worker in the 21st Century might be sustained and developed.Ways in which self organising enterprises run under co-operative principles and values might be better funded is the subject of research by Lyle Mitchell at the Open University. He asks how socially responsible investment, through innovative financial intermediaries can succeed in overcoming the financial exclusion of entrepreneurs who wish to self organise, grow and develop in mutual co-operation.

Treatment of economics from an evolutionary and institutional standpoint was suggested by the institutional economist Thorstein Veblen in the early 20th Century who thought that the human tendency towards experimentation and creative innovation could generate novelty in an ongoing manner. For Veblen, such ‘idle curiosity’ is a major source of technological change, (Hodgson-1994). His attempt to develop a dynamic theory of socio-economic evolution was based on Darwin’s synthesis but rejected genetic determinism.

Lyle recalled that Aristotle can be considered the founder of an evolutionary view of societal development and argues that viewing associations of people as complex evolving and adaptive systems was how Aristotle saw them in his political and ethical writings. Aristotle’s assumption of ‘co-operative man’ was the premise of his argument for small scale self-sufficient democratic associations where an integrated system of ethics realised the full potentials of members and justified their allegiance. His thesis questioned Plato’s proposal in the ‘Republic’ that all members of the polis should have everything in common on the grounds of impracticality. On the other hand, Socrate’s assumption that the ‘greater the unity of the state the better’ was also criticised because:

1. Such a unity would transform the polis into a household and ultimately a one man state and,

2. A polis is composed of different kinds of people exchanging their services according to their different capacities.

Aristotle therefore suggested that a sustainable association derived from a ‘principle of compensation’ was more desirable; a proportional reciprocity, where members of the polis gave to others an amount equivalent to what they received. In the ‘Nichomachean Ethics’ Aristotle refers to this as the salvation of the polis.

The reciprocity hinged upon three kinds of friendship: ‘utility’, ‘pleasure’ and

‘ virtue’. In friendships based on ‘utility’ or ‘pleasure’ a friend is not loved for what they are but from what they can provide. In the paradigmatic and primary friendship of ‘virtue’, virtuous friends are other selves to one another.

Lyle feels that with the institution of the self regulating market of the nineteenth century and the development of global capitalism, co-operative entrepreneurs have found themselves increasingly excluded from conventional forms of finance. The ‘market’ metaphor based on a reductionist and utilitarian paradigm has engendered a ‘value free and objective’ attitude which has promoted and justified the economic behaviour of individualistic profit motivated entrepreneurs. It is an attitude that assumes that entrepreneurs, consumers and producers do not have feelings , values and the potential to transform their lives but are homogenous, self interested and passive - ‘as atoms in a void responding to market signals and solely motivated by economic gain and fear of hunger..’.

It has been suggested (Blomqvist 1998,1999a and 1999b) that ‘a new practical and critical pedagogy is required where spaces for dialogue between all co-operators are made available, aiding them to become critical actors developing their own standards of excellence and virtue’.

Manoj Gamghir (?) reminded us that against the background of ‘birth, death and taxes’ there has to be something new coming along which is a variation on a previous theme. There has to be perhaps an unpleasant environment with which a system has to cope. If we have an emerging entity for which tool technology and jobs change, how do we pass on our knowledge? It may be partially coded in the tools themselves but how we train is also important. There are two aspects to this:

1. The system is reproduced by an explicit knowledge of how it works.

2. Practice is communicated by an ‘apprentice’ starting at the periphery of an organisation and going through an inductive process.

Apprentices learn from ‘masters’ though they do not have to become ‘masters’ themselves. ‘Death’ is when a ‘master’ leaves, ‘birth’ is when a novice joins. As the process proceeds so the knowledge may change. Darwinian competition may take place but where you have parts of the culture which support one another you get closure and knowledge may increase in those particular areas. In ‘leaderless’ organisations a master never teaches an apprentice but only creates the conditions by which an apprentice can best learn.

Max Boisot has worked on the problem of knowledge transfer in terms of the categorisation of information and how it diffuses as a function of the way it is abstracted from situations and codified. If codification involves drawing a line between, say logical categories a and b, how much information processing does a person do to achieve that? This is similar to the work being done by people like Gregory Chaitin who uses the notion of algorithmic complexity. A system is less codified if a description of how it is or works is short compared with the data, (or is it more?- I’m not sure-Ed.). Assuming that categories can be applied it is then a question of how many categories are needed to apprehend a particular portion of the world experience? It could be that a frog uses different categories and different numbers of categories than we do, and we might assume, though it is perhaps a fragile assumption, that if a person represents his or her experience of this room using a line drawing they are using fewer categories. As you get closer and closer to the real world you move towards a potential infinity of categories and so it becomes necessary to simplify representation, which involves for example, a move towards particle physics.

Humans try to economise on the data processing for reasons of time and energy but also to communicate and this is where the meme becomes interesting because it seems we often have a limited information space in which to fit the message. What happens is that we have certain kinds of experience which have a certain viscosity in terms of their ability to flow and then in order to communicate a particular message in a certain information space we have to make certain selections and reduce the categories. This may be easy enough with something simple but if something is complicated then memes have the ability to make things flow faster. Agents will behave differently in transactions according to whether the information exchanged is of the viscous or fluid kind. Moreover transformations which shed information which a communicator deems irrelevant begin to point at the strategies of agents.

As information moves away from the immediate experience it becomes disembodied, the context out of which the codes are precipitated are lost and the information if subsequently used has to be attached to a local context. Things become meaningful to the extent to which you can attach these codified and abstract signals to a particular or local context. When a context is not shared the communication costs go up because you have to explain the codes and the context. Max and others are trying to model the processes involved in information transfer, where the agent could either be an individual or a firm using genetic algorithms, ‘fuzzy logic’ and some mathematical modeling based on Shannon’s work in the forties. The ‘fuzzy logic’ handles the question of how you grow an alphabet. Alphabets in a language may be by agreement but there is a semantic and a pragmatic element as well as a technical one. We are beginning to see the model do interesting things; you ask it to tell an interesting story about what is seen and then see whether it’s plausible. It’s a kind of interaction between narrative and real experience.

Conclusion

It may be unwise to press the biological evolution analogy too far in studying social organisations as complex systems. Organisations are not entities in the same way that organisms are in a biosystem. It may even be more important to think of ecosystems more in terms of the ‘coral reef’ where interdependency and symbiosis is more evident than selection and competition. Should we perhaps look at an ecosystem purely in terms of its shifting webs of interdependence? We strive to understand the significance of ‘closure’ for the emergence of ‘wholes’ and our models of evolution are based on the interaction of such ‘wholes’. Is there an ontological and epistemological problem with such ‘wholes’? Can Holistic science transcend the quantitative/ qualitative and the analytical/intuitive gulf? Human organisations consist of collections of agents capable of determining their own future and whose behaviour is conditioned by the ways in which they receive information? Do memes transform the operating system or computational architecture of the human brain as Daniel Dennett suggests in his book: ‘Consciousness Explained’ (1991a, pp. 199-208)?