This book contains a collection of studies of social behaviour that are mainly biologically oriented and are carried out from the perspective of emergent effects and of self-organisation. It brings together papers that show emergent aspects of social behaviour through interaction with the environment in the entire range of organisms (from single-celled organisms via slugs, insects, fish and primates to humans). This book treats the broadest range of organisms as regards self-organisation and social behaviour that has been treated so far in one book. It is only followed by the book by Camazine et al. (2001) in which mostly insect societies are emphasised. Most of the papers deal with the direct effect of self-organisation on patterns of social behaviour. We will treat them in increasing order of complexity from slime moulds to humans (Chapters 1–8). A few papers discuss the intricate relationship between evolution and self-organisation (Chapters 9 and 10).

Before treating each of the papers in turn, a few words about self-organisation and emergent effects by interaction with the environment are needed.

Emergent phenomena arise in social systems as a consequence of self-reinforcing effects and of ‘locality’ of interactions, as explained below. Self-reinforcing effects imply that if an event takes place, it increases the likelihood that it will happen again. An example is population growth. The larger a population gets, the more individuals it contains that can bear new offspring.

The development of the social amoeba Dictyostelium discoideum

Development of a vertebrate embryo typically involves the generation of millions of cells that differentiate into hundreds of cell types to form a wide variety of different tissues and organs. Some cell types arise and differentiate in situ at the right position at the right time of development; however, many cell types have to migrate during development over considerable distances to reach their final destination. One of the best understood mechanisms guiding long-range cell movement is chemotaxis. Chemotactic cell movement is a key mechanism in the multicellular development of the social amoeba Dictyostelium discoideum. Its development is in many respects much simpler and more amenable to experimental analysis than that of vertebrates. The cells proliferate in the vegetative stage as single amoebae, which live in the soil and feed on bacteria. When the population size increases, the cells in the centre of the colony will start to starve, and starvation for amino acids acts as a signal for the cells to enter a multicellular developmental phase. Up to 105 cells aggregate to form a multicellular aggregate which transforms into a cylindrical slug. The slug migrates under the control of environmental signals such as light and temperature gradients to the soil surface, where low humidity and overhead light trigger the conversion of the slug into a fruiting body. The fruiting body consists of a stalk of dead cells supporting a mass of spores.

Introduction

In recent years statistical physics (Pathria, 1972; Ma, 1976) has been applied to a large spectrum of fields outside the scope of non-living matter (Bunde et al., 2002). While applications to social sciences are growing, they are still scarce (de Oliveira et al., 2000). In this chapter we analyse a basic ingredient of social organisations: the legitimacy of top leadership with respect to the distribution of support for various political trends present at the bottom of the organisation.

In hierarchical democratic systems each level is chosen from the one just below using a local majority rule. In principle this is supposed to yield 100% power to the larger trend. In the case of two competing trends, it means receiving more than 50% of the overall global support. This democratic ideal can seldom be satisfied, because the trend leading the organisation has several advantages. We show that accounting for such an asymmetry between the ruling trend and the challenging one may turn a democratic system into a drastic dictatorship.

This paradox is a consequence of the underlying dynamics associated with multi-level elections. It appears to obey a threshold-like dynamics, which can lead to democratic self-elimination of the huge majority against a minority trend which is in power (Galam, 1986). Indeed, repeated elections can drive the threshold for attaining power to a significant asymmetric value. For instance, it can be down to 23% for the group already in power and up to 77% for its challenging competitor.

Introduction

Perhaps no other group of organisms typifies emergent pattern as a function of the collective as well as schools of fish. Ranging in numbers from tens to millions, across all aquatic environments, trophic levels, and phylogenetic groups, fish schools are a quintessential biological model of collective action because: (1) the range of group pattern and behaviour appears to retain fundamental similarities across taxa, suggesting underlying mechanism, and (2) individuals are clearly not related to each other, as are social insects, and therefore pattern can not be simply explained by kinship altruism. Thus, schooling remains both a fundamental biological phenomenon, and a mystery. This chapter explores agent-based approaches to the study of fish schooling, with an eye towards synthesizing approaches and findings to date, and examining the degree to which synthetic results match data collected from real schools.

Specifically we will: (1) discuss emergence versus epiphenomena in the context of the individual, the group and the population; (2) review the major agent-based approaches to the study of fish schooling, with an in-depth examination of the ‘traffic rules of fish schools’; (3) compare simulation and model output to positional data collected on fish in real schools; and (4) suggest future directions for the study of fish schooling.

Emergence versus epiphenomena

We begin with a caveat: the majority of studies on fish schooling to date have centred on the question of why fish school, that is, the proximate and ultimate mechanisms behind the evolution of schooling behaviour (Pitcher and Parrish 1993).

Multiple mating in social insects

Multiple mating of females (polyandry) is a rare phenomenon in social hymenoptera (Strassmann, 2001). The adaptive value of single mating seems obvious: females should mate with as few males as possible to minimise predation risk during mating, the energy costs involved in mating and the chance of contracting a disease. Most importantly, polyandry nullifies the benefits of male haploidy for the inclusive fitness of the sterile workers in the colony. Multiple mating dramatically reduces the intracolonial relatedness (Boomsma and Ratnieks, 1996) and hence reduces the force of the arguments of kin selection theory (Hamilton 1964a, b), because the high intracolonial genotypic diversity creates an extreme potential for conflict among the nest members. Fourteen evolutionary rescue hypotheses have been identified to explain the potential benefits of polyandry in spite of the additional mating costs (Crozier and Fjerdingstad, 2001).

Although considerable effort has been expended to explain the evolution of polyandry, the consequences of polyandry for the organisation of the society have received less attention. The honeybees (Apis spp.) may be an exception in this regard. This is primarily due to a relatively large research community working on Apis because of its economic significance. The knowledge accumulated on the biology of Apis exceeds that of other social insects by far. The other rather fortunate property is the extreme degree of polyandry in the honeybee (Palmer and Oldroyd, 2000). Multiple mating in Apis can exceed 50 drones per queen (Moritz et al., 1995, Palmer and Oldroyd, 2001).

Introduction

Organisms can cope with a variable environment in which various actions are called for in a variety of ways, e.g.:

• ‘Red Queen’ evolution. Each individual performs the different types of actions with a pre-set frequency. There is some within-population/ species variation in these frequencies (i.e. it is a ‘quasi-species’ rather than a monomorphic species). Because many variants are present in the population, changes in the environment will cause relatively rapid change in the population by selection of available genotypes of the quasi-species. In such a case the rate of change is fairly independent of mutation rate.

• Frequency-dependent selection. There are two or more subtypes in the population, each specialising on one or a subset of the actions. In contrast to the previous mode the within-population variation is not unimodal but multi-modal. Dependent on which actions are more in demand these subpopulations will increase/decrease. A clear-cut example is the distribution of the ‘rover’ and ‘sitter’ types in Drosophila, which do what the names suggest in foraging. The difference has been localized to two different alleles of a cGMP-dependent kinase gene which has plural effects, among which is a change in ion channels in the brain, ultimately leading to the two modes of exploiting food resources (Osborne et al., 1997; Sokolowski, 1997; Renger et al., 1999).

• […]

The study of social systems from the perspective of complexity science leads to unusual results that show that, by self-organisation, complex patterns of behaviour may arise from very simple behavioural rules (Schelling, 1978; Camazine et al., 2001). By building these rules into certain computer models we develop a new type of understanding (Braitenberg, 1984; Pfeifer and Scheier, 1999).

This method may be applied to social systems of all kinds and of all organisms. Yet, so far, it has rarely been used among biologists. Moreover, biologists are little aware of the use of this method in the study of social systems in humans.

Therefore, we feel that there is need for a book on social systems of animals and humans from the perspective of complexity science. In order to interest also empirical scientists in this approach, the book contains both empirical papers and theoretical ones.

To realise all this a conference was essential: we organised a five-day conference in the beautiful surroundings of Monte Verità in Switzerland. The authors of the papers of this book were invited speakers at this conference. I wish to thank them for the timely submission of their papers and for their cooperative attitude during refereeing. I am grateful to Paulien Hogeweg, and Bernard Thierry for their useful general comments and to Jens Krause and Hanspeter Kunz for refereeing and to Dan Reid for his work on the index. I am grateful to the Centro Stefano Franscini of the Eidgenössische Technische Hochschule Zürich and the University of Zürich for their liberal financial support.

In their book Darwinism Evolving, Depew and Weber (1995) develop the thesis that evolutionary theory has been reformulated several times to keep pace with advances in knowledge about the physical world. Darwin's Newtonian formulation was replaced by a probabilistic formulation early in the twentieth century. According to Depew and Weber, the new science of complexity will force yet another formulation, which is taking place during our time.

This general thesis may well be correct but Depew and Weber's specific account of the relationship between evolution and complexity leaves much to be desired (Wilson, 1995). They largely accept the polemic view of Gould and Lewontin (1979) at face value, arguing that natural selection is far more constrained and adaptations less common than claimed by proponents of the so-called adaptationist programme. Complexity is viewed as something that stamps its own properties on organisms and resists the modifying effects of natural selection.

Depew and Weber are not alone in this view. Many complexity theorists and writers seem to parade under the banner ‘Darwin is dead! Long live complexity!’ The following passage by Kauffman (1993: 24) provides one example:

Introduction

The marked complexity of primate social behaviour is usually ascribed to the extraordinary intelligence of primates (Whiten and Byrne, 1986, 1997; Byrne and Whiten, 1988). Of the ‘social tools’ adopted by primates various forms of ‘bargaining’ or exchange relationship (such as the interchange of grooming for received support as supposed for chimpanzees) have drawn much attention in both cognitive and evolutionary studies. Exchange relationships are often assumed to account for the occurrence of sociopositive acts, because these acts, such as grooming, food-sharing and support in fights by primates, seem to lower the fitness of the actor and to enhance that of the receiver. The theories that are commonly applied to explain such acts are based on the assumption that the tendency to display non-selfish behaviour (so-called ‘altruism’) is genetically encoded. For each aspect, on the basis of cost–benefit arguments, separate adaptive explanations are given and separate acts are supposed to contribute independently to the fitness of an individual (so that the complete fitness of an individual equals the sum of the contributions of the separate traits). The three main theories are:

1. The kin selection theory (Hamilton, 1964). Altruism may be spreading evolutionarily if it is directed towards kin, because of the high probability that closely related individuals share the gene responsible for its realisation.

2. The theory of reciprocal altruism (Trivers, 1971). Altruism can be part of a cooperative relationship, if it is returned by the receiver to the actor. Although the altruist suffers a loss in the short term, by being ‘paid back’ later, he will benefit in the long run.

3. […]

Introduction

In animal societies, collective decisions and patterns emerge through self-organised processes, from a variety of interactions among individuals. The rules specifying these interactions are executed using only local information, that is, without reference to the global pattern. Thus collective decisions can be made that, at the individual level, require only limited cognitive abilities and partial knowledge of the environment (Camazine et al., 2001; Hemelrijk, 2002). Simple behavioural rules lead to behavioural flexibility of the society depending on its characteristics (e.g. demography, starvation and kinship) and on its environment (e.g. food distribution and presence of competitors).

Most self-organised decisions and patterns arise as a result of a competition between different sources of information that are amplified through different positive feedbacks. In contrast, negative feedbacks often arise ‘automatically’ as a result of the system's constraints (e.g. limits on the supply of food, the food reserve and the number of available workers). Amplifying communication is a characteristic of group-living animals (Deneubourg and Goss, 1989; Parrish and Keshet-Edelstein, 1999). One common type of such communication is recruitment to multiple food sources in social arthropods, but also in vertebrates and many others groups. The nature of interactions implied in these phenomena depends on the species and can involve chemical communication and/or physical contacts (Hölldobler and Wilson, 1991; Fitzgerald, 1995; Seeley, 1995; Costa and Louque, 2001; Ruf et al., 2001). Many parameters may influence the patterns of food exploitation as well as foraging efficiency.

Definition of self-organisation

Many papers that describe processes resembling self-organisation according to the definition used here do not explicitly use the term ‘self-organisation’. Instead terms like ‘emergent behaviour’, ‘population dynamics’, ‘bifurcations’, ‘catastrophes’ and others are used. In this chapter, such work will be subsumed under self-organisation. There might be slight differences in the phenomena that are described, but the basic ideas are the same, and an overview of this kind of work would be incomplete if attention was focused on only those papers that contain the term self-organisation.

Another reason why the term self-organisation is not used more frequently might be that the term itself is ill-defined. Different authors use different interpretations of the term. A selection of linguistic papers with self-organisation in the title (Lindblom et al., 1984; Wildgen, 1990; Steels, 1995; Ehala, 1996; Demolin and Soquet, 1999; de Boer, 2000; Nicolis et al., 2000) all have a slightly different view on what it is and what role it plays. Further, the term self-organisation might not be popular among linguists, because it has the negative connotation of being vague.

It is therefore useful and instructive to study in some more depth the definition of self-organisation used here. Self-organisation, according to this definition is ‘The emergence of order on a global scale through interactions on a local scale.’ The definition assumes there is a system that has two main components: actors and interactions.

As students of animal societies, we claim we observe levels of organisation, networks of relationships, mating systems and demographic structures. We identify classes, matrilines and hierarchies. We consider things like parental investment, nepotistic patterns or dominance strategies. We try to explain the patterning of these behavioural characters by looking for their fitness. Hidden in such an endeavour is a common assumption among scientists: the world and its objects exist independently of any observer; scientists have to discover these objects. Such a stance is known as metaphysical realism (Putnam, 1981). In biology, this implies that we assume natural selection to act upon characters we observe. But ‘How would we know if social organizations were not adaptive?’ (Rowell, 1979). Even if we are not acquainted with philosophical thinking, we should be warned against the appearance fallacy. We are aware that the brain reconstructs reality from electrical signals transmitted by the sensorial organs. We do not perceive social organisations per se. As Ashby (1962) puts it, the organisation exists in part in the eye of the beholder. What is seen may be named sociodemographic forms, which means, sets of individuals that are distributed and behave in a structured manner (Thierry, 1994). Sociodemographic forms represent the phenomenon, the visible aspect of social organisations. If we attempt to reduce them to adaptive strategies, this may amount to attributing an adaptive function to reified structures, in other words to endowing appearances with a fitness (Thierry, 1997).