Complex Adaptive Systems

Technical Report 070302A Complex Intelligent Systems Laboratory, Centre for Information Technology Research, Faculty of Information Communication Technology, Swinburne University of Technology Melbourne, Australia [email protected]

JASON BROWNLEE

Abstract- The field of Complex Adaptive Systems (CAS) is approximately 20 years old, having been established by physicists, economists, and others studying complexity at the Santa Fe Institute in New Mexico, USA. The field has spawned much work, such as Hollands contributions of genetic algorithms, classifier systems, and his ecosystem simulator, which assisted in provoking the fields of evolutionary computation and artificial life. The framework of inducted principles derived from many natural and artificial examples of complex systems has assisted in the investigation in such diverse fields of study as psychology, anthropology, genetic evolution, ecology, and business management theory, although a unified theory of such complex systems still appears to be a long way off. This work reviews the principles of complex adaptive systems as a framework, providing a number of interpretations from eminent researches in the field. Many example works are cited, and the theory is used to phrase some ambiguus work in the field of artificial immune systems and artificial life. The methodology of using simulations of CAS as the starting point for models in the field of biological inspired computation is postulated as an important contribution of CAS to that field. Keywords- Complex Adaptive Systems, CSA, General Principles

which highlights the diversity of such approaches.

2. Information Theory Measures of complexity and information in Hamming space as bits in terms of order and randomness. Called a theory of limits, for example cannot hold more bits than the number of synapse in our natural network.

3. Ergodic Theory The study of dissipative dynamical maps including orbits, attractors, and deterministic dynamical systems. Includes chaos theory and bifurcation theory.

4. Artificial Entities The study of artificial entities in computers such as cellular automata (CA). Such work results in simulations such as the game of

life, and the theory of edge of chaos in which complexity may be at its highest between randomness and regularity.

5. Large Random Physical Systems Systems that have statistical mechanics of complexity with complex high-dimensional attractors. These systems are typically non-ergodic such as random manifolds, percolation,

localisation, spin glass and neural networks. Also includes Kauffman s fitness landscape conception of complex systems (rather than attractors) in the context of biological evolution.

Complex Adaptive Systems (CAS) refers to a field of study and resultant conceptual framework for natural and artificial systems that defy reductionist (top-down) investigation. Such systems are generally defined as being composed of populations of adaptive agents whose interactions result in complex non-linear dynamics, the results of which are emergent system phenomena. As a field of study, CAS is concerned with (1) comparing natural and artificial examples of CAS to distil general properties and processes and (2) investigate computer simulations of simplified models of natural systems. CAS provides a conceptualisation and framework for a class complex systems and their resultant phenomena providing both computational tools and inducted principles. The field is inherently interdisciplinary, drawing strongly from complexity science, systems theory, control theory and network theory, and weakly from related fields such as statistical mechanics, artificial intelligence, game theory, and optimization. There are many ways consider complexity and to address complex systems, and the framework of complex adaptive systems is but one conceptualisation. Anderson [43] provides an insightful summary of eight popular theories or ways of thinking about complexity,

I. INTRODUCTION

6. Self-Organised Criticality (SOC) Systems in which are driven by some conservative quantity uniformly at a large scale, are able to dissipate it only to microscopic fluctuations, have fluctuations at all scales (as opposed to cycles of stability or system failure, although the systems are not considered adaptive). The systems lead to random fractals of state and scaling laws for the distributions of avalanches.

7. Artificial Intelligence (AI)

To investigate complex adaptive systems by

building them. The example provided is an expert system, Holland s adaptive systems such as genetic algorithms and classifier systems. 8. Wetware The attempt to investigate complex adaptive systems by studying them. The attempt to understand how complex systems like the brain work without attempting to specify a set of underlying principles. The naturist approach of studying systems.

Figure 1 - Summary of Anderson's eight paths to complexity theory

The study of Complex Adaptive Systems (CAS) is the study of high-level abstractions of natural and artificial systems that are generally impervious to traditional analysis techniques. Macroscopic patterns emerge from the dynamic and nonlinear interactions of the systems low-level (microscopic) adaptive agents. The emergent patterns are more than the sum of there parts, thus the traditional reductionist methodology fails to describe how the macroscopic patterns emerge. Rather, holistic and totalistic investigatory approaches are applied that relate the simple rules and interactions
page 1 of 6

COMPLEX ADAPTIVE SYSTEMS

CIS Technical Report 070302A

March 2007

of the simple adaptive agents to their emergent effects in a bottom-up manner. Generally, examples of CAS are drawn most systems studied in biology, sociology and economics. Some often cited examples include: the development of embryos, function of the adaptive immune system, ecologies, genetic evolution, thinking and learning in the brain, weather systems, market economies, trading systems, social systems, cultures, politics, traffic systems, insect swarms, the flocking of birds, implementation of new ideas, the testing of scientific theories, and bacteria becoming resistant to an antibiotic. As mentioned, computer simulated models play a large role in investigating CAS where the system is reduced to its simplest essential aspects. These simulation models themselves demonstrate the traits of complex adaptive systems and thus provide a fertile ground for controlled experimentation. Some modelling approaches used and developed for this purpose include cellular automata (CA), agent-based models (ABM), artificial neural networks (ANN), genetic algorithms (GA), and learning classifier systems (LCS).

There is no clear definition of a complex adaptive system rather sets of parsimonious principles and properties, with, in many cases, different researches in the field defining their own nomenclature [41]. This section lists some common and regarded interpretations of such general principles. In Chapter 10 of the 1992 reprint of his book [17] Holland provided a preliminary outline of the similarities and difficulties of complex adaptive systems.
Nonlinear Interactions A large number of individual parts undergoing simultaneous nonlinear interactions where the emergent behaviour is more than the some of the parts.

B.Definition by Inducted Principles

Aggregate Behaviour The impact of the system is its aggregate behaviour, the behaviour of the system as a whole, which is often feed back to the parts modifying their behaviour

Change Interaction of the parts evolves over time and the parts may face perpetual novelty. These systems typically operate far from global optimum and far from equilibrium.

Anticipation In adapting to changing circumstance, the parts anticipate the consequences of their responses. The aggregate anticipation affects the systems behaviour and this is the least understood property of such systems.

The field of complex adaptive systems was founded at the Santa Fe Institute (SFI) in New Mexico, USA, in the late 1980 s (perhaps the SFI meeting on complexity in economics in 1987, proceedings: [44]) by a group of physicists, economists, and others interested in studying complex systems in which the agents of those systems change. Perhaps one of the largest contributors to the inception of the field from the perspective of adaptation was John Henry Holland. Holland was particularly interested in adaptive systems from the perspective of genetic evolution [17]. He conceptualised an adaptive plan , which was the progressive modification of structures by means of suitable operators. From adaptive plans, he was interested in the question of how computers could be programmed so that problemsolving capabilities are built up by specifying: what is to be done rather than how to do it . A specialisation of his adaptive plan called the genetic plan ultimately contributed to the founding of the field of genetic algorithms and evolutionary computation. In the 1992 reprint of his book, he provided a summary of CAS with a computational example called ECHO. His work on CAS was expanded in a later book [16] which provided an in depth study of the topic. He also released yet another related work on theories of emergence and rules for emergent phenomena [18]. The study of complex adaptive systems was undertaking intensely at the SFI throughout the early to mid 1990 s resulting in the release of many books and papers. A few of more popular works are listed. Waldrop [38] provided a detailed review of the science of complexity, self-organisation, and adaptation recounting the history and inception of the field of CAS and its main findings. Gell-Mann [40] also produced a seminal work on complexity theory including many detailed illustrative examples. Also of note are two seminal edited volumes on CAS [9] and [39].

A. A Note on CAS History

Figure 2 - An early proposal of the general principles of CAS by Holland

Holland suggests three reasons as to why CAS may be difficult to study with conventional approaches: (1) The systems lose the majority of their features when the parts are isolated. (2) The systems are highly dependant on their history making it difficult to compare instances and derive trends. (3) They operate far from global optimum and points of equilibrium making them hard to assess with conventional approaches that are concerned with end points of systems. Holland goes on to stress the need for a unified theory of complex adaptive systems, and suggests that the framework for such a theory could be built with the mechanisms of parallelism, competition, and recombination. He also suggests that such systems respond instance-by-instant [14], and the importance of the systems ability to balance exploration (acquisition of new information or capabilities) with exploitation (efficient use of information or capabilities already available). In a detailed extension and elaboration on his contribution towards a theory of CAS [16], Holland suggests 4 properties, and 3 mechanisms, which a CAS must possess, which have become a de facto template for phrasing a system as a CAS.

Aggregation (property) Complexity emerges from the interaction of smaller components, which themselves may be the products of systems.

Tagging (mechanism) Agents are differentiated and posses a manner in which to discriminate agents with particular properties.

Nonlinearity (property) Agents interact in dynamical and non-linear ways

Flows (property) Agents organise into networks of interaction in which one interaction may trigger (flow) following interactions.

Diversity (property) Agents evolve to fill diverse niches, which are defined by the specifics of agent interactions. The concept of a niche outlives the inhabiting agents, and the evolution of niches has a larger impact on the system than the evolution of agents (levels of abstraction and control).

Internal Models (mechanism) Agents are changed through there interactions, and the changes bias future actions (agents adapt). The internal representations possess information as how to exploit the regularity of their

CIS Technical Report 070302A

March 2007

page 2 of 6

interactions, without necessarily explicitly defining that regularity. Building Blocks purposes.

(mechanism) Components are reused for multiple

Figure 3 - A paraphrase of Holland's seven aspects of a CAS

Gell-Mann [39,40] differentiates his definition from Holland s by suggesting that Holland s CAS must have a collective of interacting and adaptive agents, where his definition states that each one of Holland s agents is considered a complex adaptive system. In [39], GellMann specifies a cycle for CAS which is composed of six core elements and four concerning issues. He goes on to detail a series of questions related to investigating each point.

representation that defines agent rules and interactions. Agent fitness is optimized locally relative to the local microenvironment. The flow of information is nonlinear. Agents are tagged (perhaps heterogenous function) and aggregates of heterogenous agents can form meta-agents. Levin [52] in his phrasing of ecology and the biosphere as a CAS acknowledges Arthur s principles (as well as Holland s 4 properties), although distils them into three essential aspects.

Diversity

sustained diversity and individuality of components

Local Interactions

localised interactions between components

1. Coarse graining trade-off between the coarseness for manageability of information and fineness for adequate detail in information

Selection An autonomous process that selects from among components a subset for replication or enhancement, based on the results of local interactions amongst components Figure 6 - Three essential elements of CAS distilled by Levin from Arthur

2. Identification sorting out of regularities from randomness in information from the environment

3. Compression

perceived regularities are compressed into a schema

4. Variation

variation and improvement of schema (adaptation or evolution)

C. Features and Relations

5. Application use of schema to the systems environment, also considered decompression of schema 6. Selection the consequences of selective pressures in the real world providing feedback and affecting competition for schemata

Issues issues related to the six aspects of a CAS lifecycle include: (1) time scales, (2) the system included as a component in other systems, (3) systems with humans in the loop, and (4) system composed of many other co-adapting systems.

Figure 4 - Summary of Gell-Mann's lifecycle and related issues of a CAS

Arthur [63] proposes a definition of complex systems as studied in economics with six properties, although refers to such systems (as he claims Holland would) as adaptive nonlinear networks . These properties, like Holland s, are also cited as standard CAS principles.

Dispersed Interaction Emergent effects are the result of the actions of many dispersed, possibly heterogeneous, agents acting in parallel. The action of an agent is dependant on the anticipated actions of a limited number of other agents, and the aggregate state the limited set of agents create

No Global Controller No global entity controls interactions, rather controls are provided by competition and coordination between agents. Economic actions are mediated by the rules of the environment.

Crosscutting Hierarchical Interaction A system has many levels of organisation of and interaction. Units at any level of abstraction within the

system serve as building blocks for constructing units at high levels of abstraction. The interactions are more than hierarchical with tangled and crosscutting concerns.

Continual Adaptation behaviours, actions, strategies and products are revised continually as the individual agents accumulate experience, the system constantly adapts.

Perpetual Novelty The changes introduced by adaptation continually create new opportunities for exploration, the result is ongoing, perpetual novelty

Out-of-equilibrium Dynamics Because of the continual change and adaptation, the system operates far from global optimum and equilibrium. Improvements are always possible and regularly occur.

Figure 5

A paraphrase of Arthur's six aspects of adaptive nonlinear networks

Dooley [21] provides a condensed definition of CAS. In his definition, agents are the base elements of the system that adapt in response to interactions. The adaptations that occur operate upon agent schema, a
CIS Technical Report 070302A March 2007

Levin [53] summarises CAS and reviews some of the mathematical challenges in the field, in particular unsolved problems of pattern recognition, and the dynamics of system innovation related to the mathematics of ecology and population biology. He cautions the limited predictability of results from simulation to the natural systems. Jost [19] suggests that the environment in which a CAS exists is more complex than the CAS itself and that CAS depends on regularities in its environment. Jost goes on to provide a rigorous assessment of CAS in the context of internal and external complexity. External complexity is defined as the amount of input, information, energy the system obtains from the environment. Internal complexity is the complexity of the internal representation of the information it takes as input (model complexity). The goal of the system is to handle as much input as possible, as simple model as possible, to attempt to increase the external complexity and reduce the internal complexity of the system. Emergence is an important aspect of CAS, for example Holland devoted a book to the subject [18]. Holland s thesis was that adaptation leads to complexity, that local rules lead to emergent control and order. The trade-off in emergent control is causation at the level of the individual components of the system. SelfOrganization is another important aspect of CAS. Kauffman s thesis [59,60] is that in addition to the pressure of selection in adaptation, order can come about from self-organization, so called order for free . In his work on developing a theoretical foundation for evolutionary biology, Kauffman [60] also clearly elucidate complex systems in the context of a highdimensional fitness landscape, a notion now ubiquitous in optimization theory. The continuous adaptation in a CAS may be seen as a trade-off between too much rigidity in the face of change and too much change in the face of achieved progress (a re-phrasing of the exploration-exploitation duality). Work by Langton [3] coined the phrase edge of chaos describing complex systems as operating at critical points at the fringe (a phase transition away) from

page 3 of 6

randomness. While disputed by in the context of some experimentation (such as [36]), the theory highlights the potential fragility of these system, that a crash or failure may occur given an ineffective holistic system response. This section highlights some seminal and interesting works on both phrasing natural systems as complex adaptive systems, and computational investigations on the theory of such systems. Ecology is the study of systems of living organisms, and has proven a prototypical example of CAS. Levin [52] phrases ecology and the biosphere as complex adaptive systems in the complex of Holland s four properties, Arthur s six properties and Levin s own three distilled properties. Bonabeau [7] phrases social insect colonies (ant s, termites, etc) as CAS and show s how such systems fit all the properties discussed by Levin for ecology and the biosphere, and stresses self-organization as a critical feature in social insect systems. Hartvigsen, Kinzig, et al. [8] discuss the convention of systems theory in ecology and phrase ecosystems as complex adaptive systems. They describe the interactions in such systems being either strong and direct or weak and diffuse with positive and negative feedback. They also describe a difficulty of analysing such systems as being the large spatial and temporal scales involved. Railsback [51] applies CAS as it pertains to individual-based modelling (IBM) in ecosystems, as CAS require the modelling of the rules and interactions of individuals to obtain their emergent effects. Sigmund [20] proposes a model of reciprocal altruism in the context of CAS, investigating the evolution of cooperation with computer simulations of the prisoners dilemma (PD). Finally, Janssen [34] proposed CAS as a way to interpret global change with specific examples of the control of malaria, and climate change. In his early work mentioning CAS [17], Holland phrased a model ecosystem called ECHO as a computer simulation to investigate complex adaptive systems. This system was elaborated in is later book devoted to the topic [16]. Forrest and Jones [58] investigated CAS using ECHO. They succinctly differentiate complex adaptive systems from complex systems by the adaptive property of their agents (adopting Hollands definition). They also propose that in modelling CAS, it is desirable to strip away as much details as possible and develop models with robust behaviour (not too sensitive to parameters). They investigated broad notions of species abundance in ecosystems with ECHO with results that matched observations. Hraber, Jones, et al. [42] also investigated the abundance of species and species diversity with ECHO, commenting evolution as being a critical component of the model. Smith and Bedau [46] investigated ECHO in the context of Holland s definition of CAS. From their experimental evidence they concluded that the system lacks the diversity of hierarchically organised aggregates to be considered a CAS. Specifically the system supports the genetic diversity without the phenotypic diversity in terms of the emergent ecologies created within the system (trading and combat). They suggest an alternative to ECHO s

II. A SAMPLE OF WORKS

failure to match the definition, that perhaps Holland s definition was too specialised. They also suggest that perhaps additional features could be added to the simulation to achieve the missing effect. Finally Harris [4] provides an implementation and through treatment of the platform. Other interesting examples of natural and artificial systems phrased as complex adaptive systems include; organisation change in business (management theory) [22], supply networks [61], technology and innovation theory using patent citation rates [23], the modelling of agent-based computational economies (ACE) [30], analogy making in models of cognition (the copycat system) [35], control of the electrical power grid [1], anthropology [62] and the simulation of artificial societies [11], enterprise application integration (EAI) [12] with business objects as agents, innovation as a steady stream of novelty [15], models of the evolution of language [32], and the relationship between the management of nursing homes and quality of care [31,50]. A final interesting example is that of Goldstone and Sakamoto [45] that in the context of psychology, investigated and measured the transfer of the abstract principles of complex adaptive systems to graduate students using four different interactive scenarios. They claim that given the ubiquity of such systems, and the difficulty in grasping their inter-disciplinary principles, it is important to understand how such principles can be learned. This was in itself addressed as an example of a complex adaptive system.

The acquired immune system of vertebrates is an often cited example of a natural complex adaptive system [14,33,58]. Some examples of the acquired immune system phrased as a complex adaptive system include: A simulation of the immune system and the AIDS using a cellular automata (CA) [2]. A simulation of vaccination in the immune system using agents [6]. A general summary of the immune system and simulation approaches in the context of CAS [5]. The simulation of an immune system and HIV using elements of classifier systems, genetic algorithms and cellular automata [13]. There are many models and simulations of the acquired immune system that do not explicitly use the framework of complex adaptive system, although use some of the computational tools (such as genetic algorithms), and terminology (emergence, agents, etc.). Although the literature that matches this definition is vast, there is a pocket of work by Forrest, et al. in the early 1990 s, that does match this definition, and whose contribution is clear when interpreted with the CAS framework in mind. In summary, the work centres on a binary-encoded immune system model to study different aspects of the immune system [54] (inspired by an earlier binaryencoded pattern recognition model of the immune system [10]), which is simulated using Holland-style genetic algorithms. The system demonstrated the ability

III.COMPLEX ADAPTIVE SYSTEMS AND THE IMMUNE SYSTEM

CIS Technical Report 070302A

March 2007

page 4 of 6

to: (1) perform pattern recognition in the context of a noisy environment [57], (2) discover and maintain coverage of a diverse antigen population [47,48] (with results similar to some niching genetic algorithms for optimization), (3) to learn effectively in the presence of incomplete information [49], among other contributions. The ambiguity comes because the work has been phrased in the context of artificial life [37], and computation and pattern recognition [55]. The system was used as the basis for what evolved into computer immunology (the transitional form was a paper by Forrest, Perelson, et al. [56]) to perform pattern recognition for information technology problems such as virus detection and intrusion detection. It is this final perspective, which has blurred the work from a study of artificial life (alife), to that of artificial immune systems (AIS), which in the context of the study of complex adaptive systems, becomes quite clear. Another body of work, whose contribution is clarified in the context of complex adaptive systems from artificial immune systems, comes from Lee Segel, et al. They propose the acquired immune system as a prototype autonomous decentralised system, a prototype bottom-up artificial intelligence [24,29] (and his book on such systems from SFI [28]). As well as his conceptualisation of the immune system as an example of diffuse feedback in a diffuse system [25-27]. It is believed that these models when phrased as CAS may provoke useful starting points for systems in the field of biologically inspired computation for application to engineering and information technology problem domains.

complex biological system conforms to Holland s properties and mechanisms and Gell-Mann s CAS cycle. Conforming the immune system to this framework and to these interpretations specifically may yield additional interesting models to for exploitation in the field of artificial immune systems (AIS).

Tim Hendtlass for his patience and for providing useful feedback on drafts of this paper
REFERENCES
[1] A. M. Wildberger, Complex adaptive systems: concepts and power industry applications IEEE Control Systems Magazine, vol. 17, no. 6, pp. 77-88, 1997. IEEE Press. USA. [2] Ant nio Grilo, Artur Caetano, and Agostinho Rosa, "Immune System Simulation through a Complex Adaptive System Model ," Computational Intelligence & Artificial Life (CI & ALife 02), pp. 681-704, 2002. [3] Chris G. Langton, "Computation at the edge of chaos: phase transitions and emergent computation," Proceedings of the ninth annual international conference of the Center for Nonlinear Studies on Selforganizing, Collective, and Cooperative Phenomena in Natural and Artificial Computing Networks on Emergent computation, 1990 , pp. 12-37, 1990. [4] D. L. Harris , "Echo Implemented: A Model for Complex Adaptive Systems Computer Experimentation," Sandia National Laboratories, USA, SAND2001-2097, 2001. [5] E. Ahmed and A. H. Hashish, On Modelling The Immune System As A Complex System Theory in biosciences, vol. 124, pp. 413-418, 2006. [6] E. Ahmed, A. S. Elgazzar, and A. S. Hegazi, An Overview of complex adaptive systems ARXIV, eprint arXiv:nlin/0506059, vol. Jul, 2005. [7] Eric Bonabeau, Social Insect Colonies as Complex Adaptive Systems Ecosystems, Biomedical and Life Sciences and Earth and Environmental Science, vol. 1, pp. 437-443, Sep, 1998. [8] Gregg Hartvigsen, Ann Kinzig, and Garry Peterson, Complex Adaptive Systems: Use and Analysis of Complex Adaptive Systems in Ecosystem Science: Overview of Special Section Ecosystems, Biomedical and Life Sciences and Earth and Environmental Science, vol. 1, pp. 427-430, Sep, 1998. [9] Harold J. Morowitz and Jerome L. Singer. The Mind, The Brain, and Complex Adaptive Systems, Addison Wesley, Longman, 1995. [10] J. D. Farmer, N. H. Packard, and Alan S. Perelson, The immune system, adaptation, and machine learning Physica D, vol. 22, pp. 187-204, 1986. [11] J. Stephen Lansing, Complex Adaptive Systems Annual Review of Anthropology, vol. 32, pp. 183-204, Oct, 2003. [12] J. Sutherland and W. J. Van den Heuvel, Enterprise Application Integration Encounters Complex Adaptive Systems: A Business Object Perspective Communications of the ACM, vol. 45, pp. 59-64, Oct, 2002. [13] Joc Cing Tay and Atul Jhavar, "CAFISS: a complex adaptive framework for immune system simulation," Proceedings of the 2005 ACM symposium on Applied computing, pp. 158-164, 2005. [14] John H. Holland, Complex adaptive systems Daedalus, vol. 121, pp. 17-30, Winter, 1992. [15] John H. Holland, "Innovation in Complex Adaptive Systems: Some Mathematical Sketches," Santa Fe Institute, USA, SFI Working Paper 93-10-062, 1993. [16] John H. Holland. Hidden Order: How Adaptation Builds Complexity, USA: Addison Wesley Publishing Company, 1995. [17] John Henry Holland. Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biology, Control, and Artificial Intelligence, USA: MIT Press, 1975. [18] John Henry Holland. Emergence: From Chaos to Order, Redwood City, California, USA: Addison-Wesley, 1998. [19] Juergen Jost, External and internal complexity of complex adaptive systems Theory in biosciences, vol. 123, pp. 69-88, 2004. [20] Karl Sigmund, Complex Adaptive Systems and the Evolution of Reciprocation Ecosystems, Biomedical and Life Sciences and Earth and Environmental Science, vol. 1, pp. 444-448, Sep, 1998.

ACKNOWLEDGMENTS

Holland s perspective on complex adaptive systems has been most popular in the field of evolutionary computation, which he assisted in provoking. The genetic algorithm and learning classifier systems are staple Computational Intelligence (CI) techniques for suitable difficult domains such as design, engineering, and pattern recognition. A clear and provoking observation from this divergence, and the Forrest, et al. s immune system case study is how in both cases the adaptive models assisted to trigger related fields in CI (specifically Biologically Inspired Computation (BIC)). Undoubtedly another example would be the field of artificial neural networks as applied to classification and function approximation, having also derived from CAS models (although before CAS was conceptualised a SFI). This observation was also made by Mitchell in 1993 [33] (and I am sure others) as a viable way of contributing to CI and BIC. Perhaps the diffuse models Segal, et al. could be a case study for such an inspiration. Finally, it is important again to highlight that although CAS is an interesting and potentially potent conceptualisation, it is only that, and there has been much work in agent modelling, artificial life and computational biology (and other computational variations of physical sciences) that do not acknowledge or conform to the CAS framework. Another potentially interesting extension would be to phrase the acquired immune system as a CAS, and elucidate how this

IV. CONCLUSION

CIS Technical Report 070302A

March 2007

page 5 of 6

[21] Kevin Dooley, Complex adaptive systems: A nominal definition The Chaos Network, vol. 8, pp. 2-3, 1996. [22] Kevin J. Dooley, A Complex Adaptive Systems Model of Organization Change Nonlinear Dynamics, Psychology, and Life Sciences, vol. 1, pp. 69-97, Jan, 1997. [23] L. Fleming and O. Sorenson, Technology as a complex adaptive system: evidence from patent data Research Policy, vol. 30, pp. 1019-1039, 2001. [24] Lee A. Segel, "The immune system as a prototype of autonomous decentralized systems," IEEE International Conference on Systems, Man, and Cybernetics, 1997. 'Computational Cybernetics and Simulation, Orlando, FL, USA, pp. 375-385, 1997. [25] Lee A. Segel, Diffuse feedback from diffuse information in complex systems Complexity, vol. 5, pp. 39-46, 2000. [26] Lee A. Segel, Controlling the immune system: Diffuse feedback via a diffuse informational network Complexity in Biological Information Processing, Novartis Foundation Symposium, vol. pp. 31-51, 2001. [27] Lee A. Segel. Diffuse feedback from a diffuse informational network: in the immune system and other distributed autonomous systems. In: Design Principles for the Immune System and Other Distributed Autonomous Systems, ed. Lee A. Segel, I. R. C. New York: Oxford University Press, 2001.pp. 203-226. [28] Lee A. Segel and Irun R. Cohen. Design Principles for the Immune System and Other Distributed Autonomous Systems, USA: Oxford University Press, 2001. [29] Lee A. Segel and Ruth Lev Bar-Or. Immunology Viewed as the study of an Autonomous Decentralized System. In: Artificial Immune Systems and Their Applications, ed. D. Dasgupta. Germany: Springer, 1999.pp. 6588. [30] Leigh Tesfatsion, Agent-Based Computational Economics: Modelling Economies as Complex Adaptive Systems Information Sciences , vol. 149, pp. 262-268, Feb, 2003. [31] Lela M. Holden, Complex adaptive systems: concept analysis Journal of Advanced Nursing , vol. 52, pp. 651-657, 2005. [32] Luc Steels, "Language as a Complex Adaptive System," Proceedings of the 6th International Conference on Parallel Problem Solving from Nature, pp. 17-26, 2000. [33] M. Mitchell, Computer models of adaptive complex systems New Scientist, vol. Feb 13, 1993. [34] Marco Janssen, Use of Complex Adaptive Systems for Modeling Global Change Ecosystems, Biomedical and Life Sciences and Earth and Environmental Science, vol. 1, pp. 457-463, 1998. [35] Melanie Mitchell. Analogy-Making as a Complex Adaptive System. In: Design Principles for the Immune System and Other Distributed Autonomous Systems, ed. Lee A. Segel, I. R. C. New York: Oxford University Press, 2001.pp. 335-359. [36] Melanie Mitchell, Peter T. Hraber, and James P. Crutchfield, Revisiting the Edge of Chaos: Evolving Cellular Automata to Perform Computations Complex Systems, vol. 7, pp. 89-130, 1993. [37] Melanie Mitchell and Stephanie Forrest, Genetic Algorithms and Artificial Life Artificial Life , vol. 1, pp. 267-289, 1994. [38] Mitchell M. Waldrop. Complexity: The Emerging Science at the Edge of Order and Chaos, USA: Simon & Schuster, 1992. [39] Murray Gell-Mann. Complex Adaptive Systems. In: Complexity: metaphors, models, and reality, eds. George A. Cowan, David Pines, and David Meltzer . USA: Addison-Wesley, 1994.pp. 17-45. [40] Murray Gell-Mann. The Quark and the Jaguar: Adventures in the Simple and the Complex, Owl Books, 1995. [41] Murray Gell-Mann, What is Complexity? Complexity, vol. 1, 1995. [42] P. T. Hraber, T. Jones, and S. Forrest, The ecology of echo Artificial Life, vol. 3, pp. 165-190, 1997. [43] Philip W. Anderson. The Eightfold Way to the Theory of Complexity - A Prologue. In: Complexity: metaphors, models, and reality, eds. George A. Cowan, David Pines, and David Meltzer . USA: AddisonWesley, 1994.pp. 7-16. [44] Philip W. Anderson, Kenneth Joseph Arrow, and David Pines. Proceedings of The Santa Fe Institute Studies in the Sciences of Complexity -

Economy As an Evolving Complex System, USA: Addison Wesley Publishing Company , 1988. [45] R. L. Goldstone and Y. Sakamoto, The transfer of abstract principles governing complex adaptive systems Cognitive psychology, vol. 46, pp. 414-466, 2003. [46] Richard M. Smith and Mark A. Bedau, Is Echo a Complex Adaptive System? Evolutionary Computation, vol. 8, pp. 419-442, Dec, 2000. [47] Robert E. Smith, Stephanie Forrest, and Alan S. Perelson, "Population Diversity in an Immune System Model: Implications for Genetic Search," Proceedings of the Second Workshop on Foundations of Genetic Algorithms (FOGA), Vail, Colorado, USA, pp. 153-165, 1993. [48] Robert E. Smith, Stephanie Forrest, and Alan S. Perelson, Searching for diverse, cooperative populations with genetic algorithms Evolutionary Computation, vol. 1, pp. 127-149, 1993. [49] Ron Hightower, Stephanie Forrest, and Alan S. Perelson, "The Evolution of Secondary Organization in Immune System Gene Libraries," Proceedings of the Second European Conference on Artificial Life (ECAL '93), Brussels, Belgium, 1993. [50] Ruth A. Anderson, L. Michele Issel, and Reuben R. McDaniel Jr. Nursing homes as complex adaptive systems: relationship between management practice and resident outcomes Nursing Research, vol. 52, pp. 12-21, Jan, 2003-Feb 28, 2003. [51] S. F. Railsback, Concepts from complex adaptive systems as a framework for individual-based modelling Ecological Modelling, vol. 139, pp. 47-62, Mar, 2001. [52] Simon A. Levin, Ecosystems and the Biosphere as Complex Adaptive Systems Ecosystems, Biomedical and Life Sciences and Earth and Environmental Science, vol. 1, pp. 431-436, Sep, 1998. [53] Simon A. Levin, Complex adaptive systems: Exploring the known, the unknown and the unknowable American Mathematical Society, vol. 40, pp. 3-19, 2003. [54] Stephanie Forrest and Alan S. Perelson, "Genetic Algorithms and the Immune System," Proceedings of the 1st Workshop on Parallel Problem Solving from Nature, Dortmund, Germany, pp. 320-325, 1991. [55] Stephanie Forrest and Alan S. Perelson, Computation and the immune system ACM SIGBIO Newsletter, Special edition on biologically motivated computing, vol. 12, pp. 52-57, Jun, 1992. [56] Stephanie Forrest, Alan S. Perelson, Lawrence Allen, and Rajesh Cherukuri, "Self-Nonself Discrimination in a Computer," Proceedings of the 1992 IEEE Symposium on Security and Privacy, pp. 202-212, 1994. [57] Stephanie Forrest, Brenda Javornik, Robert E. Smith, and Alan S. Perelson, Using Genetic Algorithms to Explore Pattern Recognition in the Immune System Evolutionary Computation, vol. 1, pp. 191-211, 1993. [58] Stephanie Forrest and Terry Jones, Modeling Complex Adaptive Systems with Echo Complex Systems: Mechanism of Adaptation, vol. 2, pp. 3-21, 1994. [59] Stuart A. Kauffman. At Home in the Universe: The Search for Laws of Self-organization and Complexity , USA: Oxford University Press, 1995. [60] Stuart Kauffman. The Origins of Order: Self-Organization and Selection in Evolution, New York, USA: Oxford University Press, 1993. [61] Thomas Y. Choi, Kevin J. Dooley, and M. Johnny Rungtusanatham, Supply Networks and Complex Adaptive Systems: Control Versus Emergence Journal of Operations Management, vol. 19, pp. 351366, 2001. [62] Tom Abel, Complex Adaptive Systems, Evolutionism, and Ecology within Anthropology: Interdisciplinary Research for Understanding Cultural and Ecological Dynamics Georgia Journal of Ecological Anthropology, vol. 2, pp. 6-29, 1998. [63] W. Brian Arthur. Introduction: Process and Emergence in the Economy . In: The Economy as an Evolving Complex System II, eds. W. Brian Arthur, Steven Durlauf, and David A. Lane. Reading, Mass, USA: Addison-Wesley Pub. Co, 1997.pp. 1-04.

CIS Technical Report 070302A

March 2007

page 6 of 6