View sample Models Of Emergent Social Behavior Research Paper. Browse other social sciences research paper examples and check the list of research paper topics for more inspiration. If you need a religion research paper written according to all the academic standards, you can always turn to our experienced writers for help. This is how your paper can get an A! Feel free to contact our research paper writing service for professional assistance. We offer high-quality assignments for reasonable rates.
By the term ‘emergent social behavior’ we mean those patterns of social behavior that arise automatically without relevant speciﬁcations being in-built in the individual. Under what conditions such patterns may emerge, and of what kind they are, is unpredictable in itself, but can be learned from ‘bottom-up’ models. In such models, a so-called ‘process-oriented’ approach is used. An artiﬁcial system is constructed in which agents are equipped with behavioral rules and mechanisms. While these agents interact with their environment and other agents, their behavior is studied in the same way as in ethological studies. Results show that in this way simple agents generate complex social structures, and that the same set of rules may lead to diﬀerent patterns depending on the past experiences of the individuals, the demography of their population or the distribution of their food.
Need a Custom-Written Essay or a Research Paper?
Academic Writing, Editing, Proofreading, And Problem Solving Services
Studies of such complex systems were, of course, made possible by the development of computers, which have led to new modeling techniques. Initially, these studies dealt with emergent phenomena in chemical and physical research (Gleick 1987). In this research paper, the major issues and characteristics of these models (called ‘individual-oriented,’ ‘individual-based,’ or artiﬁcial life models) are explained by examples. Their connection with views on evolution and their general usage is discussed.
1. Major Issues Of Self-Organized Social Behavior
Computer models in question deal with the emergence of:
(a) the formation of groups;
(b) collective ‘decisions’ (e.g., as regards the choice of a food source);
(c) the distribution of tasks over individuals (such as brood care and foraging);
(d) a reproductive ‘clock’;
(e) spatial structure of groups; and
(f ) social organization of groups.
1.1 Group Formation
Many animal species live in groups which vary in characteristics, such as composition, size, permanency, and cohesion. One of the most cohesive and largest societies is found in army ants (with populations of up to 20,000,000 ants). Remarkably, these ants are practically blind and rely on a pheromonal system with which they mark their paths and by which they follow paths taken by others. Preying upon insect colonies, they raid over the forest in phalanxes of about 200,000 workers. Diﬀerent species display different swarming patterns. To explain the diﬀerences among species in swarming, it is usually assumed that there are corresponding diﬀerences in the underlying behavioral motivations. Deneubourg and co-authors (1989), however, have shown that such markedly diﬀerent swarming patterns may arise from one and the same system of laying and following pheromone trails, when ants are raiding in diﬀerent environments with diﬀerent distributions of food. To show this, the authors used a simulation in which ‘artiﬁcial ants’ moved in a discrete network of points, and marked their path with pheromones. Further, when choosing between left and right, they preferred the more strongly marked direction. By introducing diﬀerent distributions of food, diﬀerent swarming patterns arose from the interaction between the ﬂow of ants heading away from the nest to collect food, and the diﬀerent spatial distributions of the foragers returning with the food. These diﬀerent swarm types were remarkably similar to those of certain army ants.
1.2 Collective ‘Decisions’
Further, the use of trail pheromones is also a mechanism of collective ‘decision’ making. Experiments with real ants combined with modeling have shown that trail laying often causes ants to choose the food source that is nearest to the nest (Deneubourg and Goss 1989). This observation seems to suggest that individual ants possess considerable cognitive capacities, but Deneubourg and Goss show in a computer model that ants may accomplish this remarkable feat without comparing distances to various food sources. When ants mark their path by pheromones and follow the more strongly marked branch at crossroads, then, by returning to the nest more quickly when the road to and from the food source is shorter, they obviously imprint the shorter path more heavily with pheromones. Thus, positive feedback is the result: as the shorter path attracts more ants, it also receives stronger marking, etc.
In subsequent extensions of this model (Detrain et al. 1999), path marking and following are shown to imply automatically risk avoidance (which agrees with observations in some species of ants), as follows: a hostile confrontation delays the forager for a certain time before returning to the nest. Thus the path to the dangerous food source automatically is marked more slowly and less strongly. Consequently, the safest food source is usually ‘preferred.’
1.3 Distribution Of Tasks
Such a division of labor occurs in colonies of many species of social insects. Remarkably, such division is ﬂexible, i.e., the ratios of workers performing diﬀerent tasks may vary according to the changing needs of the colony. The earliest model on the emergence of task division is that designed for bumblebees (Bombus terrestris) by Hogeweg and Hesper (1983, 1985). In an earlier experimental study, it was discovered that, during the growth of the colony, workers developed into two types, ‘common’ and ‘elite’ workers. The activities carried out by the two types diﬀer remarkably: whereas the ‘common’ workers mainly forage and take rests, the ‘elite’ workers are more active, feed the brood, interact often with each other and with the queen, and sometimes lay eggs. To study the conditions for the formation of the two types of workers, Hogeweg and Hesper (1983, 1985) set up a so-called ‘Mirror’ model. This kind of model is complex in the number of variables included, but simple as regards behavioral processes. It contains biological data concerning the time of the development of eggs, larvae, pupae, etc. Space in the model is divided into two parts, peripheral (where inactive common workers doze part of the time) and central (where the brood is, and where all interactions take place). The rules of the artiﬁcial adult bumblebees operate ‘locally’ in that their behavior is triggered by what they meet. What they meet is picked out in the model randomly from what is available in the space in which the bumblebee ﬁnds itself. For instance, if an adult bumblebee meets a larva, it feeds it, and if it meets a pupa of the proper age, it starts building a cell in which a new egg can be laid, etc. All workers start with the same dominance value after hatching, with only the queen starting with a much higher rank. When one adult meets another, a dominance interaction takes place, the outcome of which (victory or defeat) is self-reinforcing. The positive feedback is ‘damped,’ however, because expected outcomes reinforce or diminish the dominance values of both partners only slightly, whereas unexpected outcomes give rise to relatively large changes in dominance value. Dominance values of the artiﬁcial bumblebees inﬂuence almost all their behavioral activities; for instance, individuals of high rank are less likely to forage.
It appears that this model generates two stable classes automatically, the ‘elites’ and ‘commons,’ with their typical dominance and behavioral conduct. For this result, the nest has to be divided into a center and a periphery (as found in real nests).
The ﬂexibility of the distribution of tasks becomes obvious upon halving the worker force in the model. In line with observations in similar experiments on real bumblebees, this increases markedly each individual’s tendency to forage. This arises in the model as follows: the lowered numbers of workers reduces the number of encounters among them, and thus increases the frequency of encounters between bumblebees and brood. Encountering the brood more often induces workers to collect food more frequently.
A diﬀerent approach to the eﬀect of social interactions on task allocation was taken by Pacala et al. (1996). In their model, individuals will stick to a certain task as long as it can be executed successfully, but otherwise give up (e.g., stop foraging when food is depleted). Then, as soon as they encounter another individual that is performing a task successfully, they will switch to perform that task too.
Another approach to the emergence of task division in social insects is described by Bonabeau et al. (1996). Their basic assumption is that diﬀerent types of workers have diﬀerent, genetically determined response thresholds for behavioral tasks. This is particularly relevant for dimorphic species, such as the ant Pheidole megacephala, in which two physically diﬀerent castes, the so-called majors and minors, exist. Majors generally have higher thresholds than minors.
This ﬁxed threshold model has been extended to variable response-thresholds by Theraulaz et al. (1998). They added learning in the form of a reinforcement process. This implies that performing tasks lowers the threshold of task execution and not- performing them heightens it.
1.4 Behavioral ‘Clock’
In real bumblebees, the queen switches at the end of the season from producing sterile oﬀspring to fertile ones, after which she is either chased away or killed in spite of the fact that she is still as dominant as before. The time of departure death of the queen is important for the ‘ﬁtness’ (i.e., number of new queens) of the colony.
After the demise of the queen, new queens and males (so-called generative oﬀspring) are reared. It is assumed that the switch should take place some weeks before the end of the season, because at that time the colony is at its largest and can take care of the largest number of oﬀspring. But it is very diﬃcult to think of a possible external signal that could trigger such a switch. However, Hogeweg and Hesper (1983) show in their Bumblebee model that no external signal is needed, and that the switch may occur as a socially regulated ‘clock’ involving a ﬁxed threshold, as follows. During the development of the colony, the queen is able to inhibit worker ovipositions by producing a certain pheromone that delays their production of eggs. Just before she is killed, she can no longer suppress the ‘elite’ workers, because, due to the large size of the colony, individual workers are less subjected to the dominance of the queen. So they start to lay unfertilized (drone) eggs. Further, the stress on the queen increases each time she has to perform dominance interactions with workers during her egg laying. When the ‘stress’ of the queen is at a low ﬁxed threshold value, she switches to producing male eggs (drones), and when it increases to a high ﬁxed threshold, she is killed or chased away.
The notion of this switch, which functions as a kind of social clock, has been challenged, because generative oﬀspring are sometimes also found in small nests. However, in response to this objection, Hogeweg and Hesper have shown that the switch in small nests created in the Bumblebee model by reducing the growth speed, appears to occur at the same time as in large nests (1985). This is due to a complicated feedback process.
1.5 Spatial Structure
In many animal groups, there is a spatial structure in which dominants are in the center and subordinates at the periphery. This spatial structure is usually explained by the well-known ‘selﬁsh herd’ theory of Hamilton (1971). The basic assumption of this theory is that individuals in the center of a group are best protected against predators, and therefore have evolved a preference for positions in which group members are between them and the predators, the so-called ‘centripetal instinct.’ However, such a spatial structure with dominants in the center also emerges in a model designed by Hemelrijk (2000), called DomWorld, in which such a preference for the center is lacking. The model consists of a homogeneous virtual world inhabited by agents that are provided with only two tendencies: (a) to group, and (b) to perform dominance interactions. Dominance interactions reﬂect competition for resources (such as food and mates) and the agent’s capacity to be victorious (i.e., its dominance) depends, in accordance with ample empirical evidence, on the self-reinforcing eﬀects of winning (and losing).
In this model, the spatial conﬁguration arises without any ‘centripetal instinct,’ but is the result of the mutually reinforcing eﬀects of spatial structure and strong hierarchical diﬀerentiation. It can be explained as follows. Pronounced rank diﬀerentiation causes low-ranking agents to be chased away by many, and therefore to end up at the periphery. Automatically, this leaves dominants in the center. Besides, this spatial structure causes agents to interact mainly with partners close in rank; therefore, if a rank change occurs, it is only a minor one. Thus, the spatial structure stabilizes the hierarchy and maintains its diﬀerentiation.
Spatial structure in human organizations is, for instance, evident in ghetto formation and in the formation of same-sex groups at parties. The Harvard economist, Thomas Schelling (1971), has shown how a stronger segregation emerges than is wanted by the individuals—who have only a slight preference for their own type—due to a positive feedback. Once individuals obtain minority status in a subgroup, because they are surrounded by an insuﬃcient number of individuals of their own type (race or sex), they leave the subgroup, and this induces others of the same type to leave their former subgroup too. Their arrival in another subgroup may in its turn induce individuals of the other type to leave. Accordingly, Schelling notes that, even if the preference to have some individuals of one’s own type close by is slight, its behavioral consequences may increase at a group level, leading to unexpected ‘macropatterns.’
1.6 Social Organization Of Groups
Two types of society are generally distinguished in the study of animal behavior: egalitarian and despotic ones. In certain primate species, such as macaques, a steep dominance or power hierarchy characterizes a despotic society, and a weak hierarchy an egalitarian one. Furthermore, both societies diﬀer in a number of other characteristics, such as bidirectionality of aggression and cohesion of grouping. For each of these (and other unmentioned) diﬀerences a separate, adaptive (i.e., evolutionary) explanation is usually given, but DomWorld provides us with a simpler explanation for many of them simultaneously (Hemelrijk 1999): by simply increasing the value of one parameter, namely, intensity of aggression (in which egalitarian and despotic macaques diﬀer—see Thierry 1990), the artiﬁcial society switches from a typical egalitarian to a despotic society with a cascade of consequences that resemble characteristics observed in the corresponding society of macaques. Such resemblances imply that diﬀerences between egalitarian and despotic societies of macaques may be entirely due to one trait only, intensity of aggression. On an evolutionary timescale, one may imagine that species-speciﬁc diﬀerences in aggression intensity arose, for example, because in some populations of a common ancestor, individuals suﬀered more food shortages, and consequently only those with intense aggression could survive.
Apart from variation among individuals in power, human societies are also characterized by a distribution of wealth. Epstein and Axtell (1996) show how, under many circumstances, a very skewed distribution of wealth emerges among heterogeneous agents (for instance, diﬀering in vision and metabolism) in a model called ‘Sugarscape.’ Such a skewed distribution of wealth appears to be a general outcome of many speciﬁcations of agents that are extracting resources from a landscape of ﬁxed capacity.
2. Characteristics Of Models For Emergent Social Patterns
Typical of models for self-organized social behavior are the following characteristics of some or all of the above examples: (a) nearby perception and activation, (b) positive (instead of the more commonly used negative) feedback processes, (c) spatial distribution, (d) the possibility of studying complex behavior, and (e) ‘open-endedness’.
(a) Nearby perception and activation. Individual agents have exclusive information about what is happening nearby. Further, timing of acts of agents is not ruled by a central clock, but nearby agents activate each other. Otherwise, agents are activated randomly.
(b) Positive feedback processes. Examples of these are the trail marking (with pheromones) and tracking system of ants, winning and losing of dominance interactions and a mutual reinforcement between spatial structure and the development of the hierarchy among group-living agents.
(c) Spatial distribution. The inclusion of space in the model creates the possibility for spatial structures to emerge. Such a spatial structure may aﬀect behavior of individuals.
(d) The possibility of studying complex Behavior. In these models, behavior is recorded in behavioral units similar to those used by ethologists for studying the behavior of real animals. Data are analyzed to detect new behavioral activities and patterns.
(e) ‘Open-endedness’—that is, at the time when the model is designed, it is unknown what patterns may emerge.
3. Emergence And Evolution
Because in these models behavioral mechanisms are implemented without continuous reference to pay-oﬀ beneﬁts, this procedure leads to a parsimonious or restrictive view of evolution in the sense that the behavioral patterns observed need not always be based on trait-speciﬁc genes, and are not willed expressly by the individuals. Besides, many of the behavioral patterns usually attributed to long-term evolution may instead be explained as direct, short-term eﬀects of other mechanisms. In these respects, these models agree with the social genetic and evolutionary explanations of, for instance, Lewontin, Gould, and Eldredge (Depew and Weber 1995), and they contrast with the thermodynamic and gene-focused views of Fisher and of sociobiology. In sociobiology, it is not the ‘processes’ that are of interest, but questions regarding adaptivity of behavior, and regarding evolutionarily stability of strategies. These issues are studied in the sociobiologically-oriented gametheoretic models. In contrast to models for emergence, behavioral strategies (such as ‘hawks and doves,’ ‘cooperators and defectors’) are deﬁned in terms of pay-oﬀ without taking into account the underlying mechanisms or behavioral and environmental features; entities with diﬀerent strategies are pitted against each other in order to examine which maintain themselves and which disappear, or in what proportions they may continue to exist in combination; new strategies that are not included from the start cannot arise in these game-theoretic models.
4. The Beneﬁts Of Models For Emergent Social Behavior
Models of self-organization are never to be regarded as a proof of how behavior arose. They are, however, useful for several reasons. First, they may suggest explanations that cannot be reached by theorizing alone. Take, for instance, Camazine’s model on the structuring of the comb of honey bees (Camazine 1991). The comb of honey bees is structured with brood in the center, and pollen and honey in layers around it. This organization is usually attributed to a cognitive blueprint in the honey bees. Camazine’s model shows, however, that the structuring of the comb may also result as a side-eﬀect from the feeding activities of the brood, and from the relative quantities of pollen and honey imported or removed.
Second, these models can be used as ‘search indicators’ for empirical studies. If models are ﬁrmly based on observed biological mechanisms (as in many of the above mentioned examples), and if the emerging patterns agree with the patterns observed in real groups of organisms, then additional patterns observed in the model, but not yet noted in the real world, may be looked for in the real world. In the same way, in a model that has been shown to be relevant for real animal behavior, consequences of certain parameters may also be studied, e.g., the eﬀects of diﬀerent degrees of cohesion of grouping. To carry out such experiments in computer simulation is very easy. Usually, similar experiments are much more diﬃcult in the real world.
Third, the insight that certain patterns of social behavior may be emergent often leads to new, parsimonious hypotheses about their evolution.
Fourth, such models can also be used in teaching pupils how to think in terms of self-organization in general as is done, for instance, by Resnick (1994).
Fifth, certain kinds of insight obtained from models may be used to solve real-world-problems (e.g., Bonabeau and Theraulaz 2000). For instance, the ant models by Deneubourg have been extended by Dorigo to solve the famous ‘traveling salesman problem.’ In this problem, a person must ﬁnd the shortest route by which to pay a single visit to a number of cities. This problem is very diﬃcult to solve, because the possibilities easily run into the millions.
- Bonabeau E, Theraulaz G 2000 Swarm smarts. Scientiﬁc American 282: 72–80
- Bonabeau E, Theraulaz G, Deneubourg J-L 1996 Quantitative study of the ﬁxed threshold model for the regulation of division of labour in insect societies. Proceedings of the Royal Society London B: Biological Sciences 263: 1565–9
- Camazine S 1991 Self-organizing pattern formation on the combs of honey bee colonies. Behavioural Ecology and Sociobiology 28: 61–76
- Deneubourg J L, Goss S 1989 Collective patterns and decision-making. Ethology, Ecology and Evolution 1: 295–311
- Deneubourg J L, Goss S, Franks N, Pasteels J M 1989 The blind leading the blind: modeling chemically mediated army ant raid patterns. Journal of Insect Behavior 2: 719–25
- Depew D J, Weber B H 1995 Darwinism Evolving. Systems Dynamics and the Genealogy of Natural Selection. MIT Press, Cambridge, MA
- Detrain C, Deneubourg J-L, Pasteels J M 1999 Decision-making in foraging by social insects. In: Detrain C, Deneubourg J-L, Pasteels J M (eds.) Information Processing in Social Insects. Birkhauser Verlag, Basle, Switzerland, pp. 331–54
- Epstein J M, Axtell R 1996 Growing Artiﬁcial Societies. Social Science from Bottom Up. MIT Press, Cambridge, MA
- Gleick J 1987 Chaos: Making A New Science. Viking, New York
- Hamilton W D 1971 Geometry for the selﬁsh herd. Journal of Theoretical Biology 31: 295–311
- Hemelrijk C K 1999 An individual-oriented model on the emergence of despotic and egalitarian societies. Proceedings of the Royal Society London B: Biological Sciences 266: 361–9
- Hemelrijk C K 2000 Towards the integration of social dominance and spatial structure. Animal Behaviour 59: 1035–48
- Hogeweg P, Hesper B 1983 The ontogeny of interaction structure in bumble bee colonies: a MIRROR model. Behavioral Ecology and Sociobiology 12: 271–83
- Hogeweg P, Hesper B 1985 Socioinformatic processes: MIRROR modelling methodology. Journal of Theoretical Biology 113: 311–30
- Pacala S W, Gordon D M, Godfray H J C 1996 Eﬀects of social group size on information transfer and task allocation. Evolutionary Ecology 10: 127–65
- Resnick M 1994 Turtles, Termites and Traﬃc Jams. Explorations in Massively Parallel Microworlds. MIT Press, Cambridge, MA
- Schelling T C 1971 Dynamic models of segregation. Journal of Mathematical Sociology 1: 143–86
- Theraulaz G, Bonabeau E, Deneubourg J-L 1998 Threshold reinforcement and the regulation of division of labour in social insects. Proceedings of Royal Society London B: Biological Sciences 265: 327–33
- Thierry B 1990 Feedback loop between kinship and dominance: The macaque model. Journal of Theoretical Biology 145: 511–21