Assessment of Complex Adaptive System Changeability Using a Learning Classifier System

Complex Adaptive Systems (CAS) are ubiquitous and composed of many interacting “agents” that exhibit independent properties and behaviors that function together with their environment to produce emergent properties. Emergence and emergent properties cannot be predicted by isolated understanding of these interacting agents/components; but can be demonstrated by observing the outcomes associated with dynamically changing interacting components of a CAS. Obviously, evolution also plays a key role in driving emergence as a defining feature of biological CAS. Biological CAS are highly heterogeneous and complex, both within and across broad scales of time and space. Biological CAS are also non-linear which means that predicting outcomes is difficult. Moreover, it is impossible to “fix” a CAS, rather the identification of leverage points that can alter the trajectory of the system to achieve desired outcomes is a more logical approach. Until recently, systems of such high dimensionality were not sufficiently tractable to understand and apply CAS principles to a disease as complex as cancer. However, recent progress in the development of advanced technologies such as computation, machine learning, artificial intelligence, and modeling portend a day when cancer will be viewed and managed as a CAS. These “big data” tools offer new and innovative opportunities to mine, manage, manipulate, model and simulate cancer to derive the information needed to manage it as a CAS. In terms of applying these principles at least one approach to treating cancer, immunotherapy, suggests that achieving a future state where cancer is viewed through the lens of CAS is well underway. Immunotherapy represents a paradigm shift in cancer therapy in that it targets the immune system, which is a quintessential CAS. When immunotherapy is successful the outcome is a “homeostatic reset” of what is an extraordinarily complex interaction between cancer and the immune system. Together these two complex systems comprise a CAS that promises to re-define how we treat and prevent cancer. A variety of immunotherapeutics (dominated by checkpoint inhibitors) have produced durable responses (possible cures) in a few patients against some cancers. These agents essentially block signals that the tumor employs to keep the immune system from recognizing and killing the cancer. However, the interaction of cancer and the immune system is a dynamic CAS that will ultimately require a detailed understanding of the cellular and microenvironmental changes that occur in patients in response to specific immunotherapeutic interventions. The challenges we face are significant including: identifying responders/non-responders; determining doses; predicting and controlling toxicities; developing rational combinations; and creating new targeted systems-based therapies. Fortunately, many of these challenges can be met by defining the “states” produced by some of the defining alterations observed in responsive and non-responsive patients including “omics” alterations, types of immune cells, temporal relationships, immune activation, humoral factors, etc. Although early, models and platforms to describe, annotate, model and simulate these systems alterations are emerging. In the past several years, we have developed a modeling platform that permits the study of the immune system and its interaction with cancer. “Cell Studio” is an immune-modeling engine that seeks to examine cancer and the immune response as a dynamic CAS by using real world data on the immune system and cancer to develop and inform computational models. Cell Studio permits the user to conduct in silico experiments of defined time and complexity. It combines agent-based and mathematical modeling approaches to capture multiscale dynamics within the immune system. The engine permits user creation of multiple different types of immune cells each with different classes of properties including different collections of cell surface receptors at different concentrations and affinities as well as the capacity to release and respond to cytokines. Multiple compartments corresponding to different body niches (e.g. lymph node, tumor environment) can be created. Mathematical models govern phenomena such as diffusion and cell tracking of cytokine gradients. As a CAS, based on a finite number of “rules” the system is self-organizing and can display emergent properties. User defined therapeutic interventions such as drug administration can be incorporated to assess the system’s response. Cell Studio is implemented using a gaming platform so that the in silico experiments can be visualized in 3D - in real time if desired. This permits researchers to perform experiments similar to those done using biologic model systems and visualizing the results. Like most video gaming platforms, different user views, overviews, individual cell movement, etc. are available and real-time as well as cumulative statistical outputs are captured and displayed. Unlike biologic model systems the simulations can be time reversed to identify, visualize, and manipulate key events. Experimentation using the Cell Studio modeling engine shows that it can recapitulate the longitudinal events in biologic model systems. Additionally, it can recover “immunophenotypes” observed in human studies of immunotherapy in cancer. It is anticipated that the in silico modeling can augment current biologic modeling strategies - especially since it can be run with numbers of replicates of virtual experiments that are not practical with biologic model systems. Additionally, it promises to assist in the rationale to develop combinatorial interventions hitting multiple immune targets and in understanding factors that modulate successful outcomes. In summary, the implications of viewing, studying and developing strategic approaches to fundamentally understand the cancer - immune system CAS are profound. Cell Studio is a next generation novel and powerful approach to analyze and model specific components of this dynamic and integrated CAS for the benefit of patients. Citation Format: Anna D. Barker, Kenneth Buetow. Viewing cancer as a complex adaptive system and managing immunotherapy as “homeostatic reset” [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr SY36-03.

Complex adaptive systems (CAS), to reiterate, are systems composed of many individual parts or agents in which patterns can emerges as a result of agents deploying "simple rules" from the "bottom-up" without external control—CAS are "self-organizing" systems. "Simple rules" in health care would include seeking to optimize both patient well-being and the functioning of professionals. If elements of a CAS system are altered, the system adapts or reacts. The behaviour of a complex adaptive system can be inherently unpredictable and non-linear as elements of the system, the internal (eg, professionals and managers) and external agents (eg, patients, families, and society), have multiple perturbations, changes, and interdependencies. Despite the flurry of interest in complex systems and non-linear dynamics in recent decades, application of knowledge and innovation about complexity and adaptation in systems for health care has been slow. Critics typically state that there is no "evidence" that applying CAS and complexity science is needed or "works" in the real world of health care systems.1 It is almost a decade since the issues of practicability were first raised in this Forum in 2009.2 Has progress been made? A PubMed scan (Figure 1) provides some comfort in the growth of applications of CAS thinking in health research. In this Forum, Wietmarschen, Wortelboer, and van der Greef3 provide a highly accessible vision for the future of complex adaptive systems and why they are needed. They re-articulate why a shift is needed from static silos of diagnoses and linear structures toward a more integrated biopsychosocial way of thinking about health, using systems thinking approaches. Moreover, in their far-sighted appraisal of Western lifestyle problems of obesity and sedentary behaviours, they demonstrate practical modelling techniques integrating molecular with cognitive and psychological metrics, and variables from different layers of human functioning. A systems dynamics software tool called Method to Analyse Relations between Variables using Enriched Loops was used to create the model during the group sessions. The resulting model contained various positive and negative feedback loops connecting multiple health domains, indicating non-linear mechanisms affecting processes that cross multiple health domains. These techniques have been applied to the analyses of individual trajectories in a clinical approach to obesity in Vogellanden-Centre for Rehabilitation, Zwolle, the Netherlands. System dynamics modelling (SD), is an interdisciplinary modelling method used for representing and understanding the behaviour of complex systems. An SD model consists of a series of stocks, which represent the total people receiving a type of service at a given time, interconnected through flows, which represent the movement of people from one stock to another over time. Participatory approaches align stakeholder understanding of the underlying causes of a problem and can achieve consensus for action. Advances in software are allowing the participatory model building approach to be extended to more sophisticated multimethod modelling that provides policy makers with more powerful tools to support the design of targeted, effective, and equitable policy responses for complex health problems.4 Cepoiu-Martin and Bischak5 utilized a system dynamics model of the Alberta Continuing Care System (ACCS), Canada, using stylized data to assist service planning. They explored policies of introducing staff/resident benchmarks in both supportive living and long-term care (LTC) in the background of predicted increases in the population of people with dementia and the provision of staffing benchmarks, The ACCS model developed, by going beyond linear cause-effect considerations, and allowed the exploration of the entire network of causal relationships between various components of the system. It provided evidence of applicability of SD simulation to analysis of the impact of adopting benchmarks related to the staff/resident ratios in the continuing care system in Alberta. The model provides a basis for future evaluations of interventions in the workforce development area, capturing all feedbacks that modify balance between staff supply and demand in the age care sector. The following three papers highlight practical applications at the clinical coal-face, albeit all are early stage studies. Bandini et al6 have successfully piloted a clinical tool for episode complexity in inpatient care on internal medical wards. Episode complexity represents the need for greater time and effort (compared with other patients and episodes) with respect to clinical assessment and treatment; relationships with the patients, caregivers, other specialists, and actors in the health care network; and information gathering and processing. A very interesting emergent finding from their study is that multimorbidity as measured by the Charlson comorbidity index was not a good predictor of episode complexity, as patients with multiple comorbidities often had simple hospital episodes while those without little comorbidity (low Charlson comorbidity score) had much more complex episodes with much less certain outcomes. The dynamics of those individual illness trajectories were not predicted by standard static disease based metrics nor supported by guidelines. Individual trajectories or journeys is a recurring theme in the developing CAS approaches in health care, representing the opportunity for responding to health status dynamics in a timely manner.7 This notion of intellectual work and time as markers of clinical complexity was also raised by Katerndahl et al in a previous analysis of medical work across clinical specialities.8 In complex systems, as the information in the input increases linearly, the complexity of the system increases exponentially. Thus, a simple rule is suggested, that clinical work complexity reflects the amount of care provided weighted by its diversity and variability. Primary care, because of its diversity and variability, scores highly on the amount of work demanded of its practitioners. In this theme, Fink et al9 describe the application of a clinical tool—Diagnostic Protocols (DP)—in a single handed practice over a 14-year period. Based on several decades of work by Braun and colleagues, DP represents a series of simple rules to reduce uncertainty in primary care presentations of serious conditions that may seem at first contact to be routine and non-serious. Here, we have the common theme of simple rules to identify courses of action related to simple and complex dynamics in patient trajectories over time in clinical care. At an organizational level, leadership is a crucial element of success, and its role is recognized as an important factor for achieving better performance and optimizing health improvements for patients. Horvat and Filipovic,10 using complexity leadership theory, identified three types of leadership and matched them to indicators of organizational maturity. Administrative leadership is grounded in traditional, bureaucratic notions of hierarchy, alignment, and control. Enabling leadership structures and enables conditions in which CAS can optimally address creative problem solving, adaptability, and learning. Adaptive leadership exemplifies a generative dynamic that underlies emergent change activities. Organizational maturity promotes organizational learning, enables effective and efficient management performance, reduces errors, and adapts to internal and external dynamics. Sustained success can be achieved by the effective management of the organization, through awareness of the organization's environment, by learning, and by the appropriate application of either improvements, or innovations, or both.11 Their survey of Serbian managers supported the hypothesis that administrative leadership had little influence on any maturity category of health care organizations. Adaptive and enabling leadership had greater association with managerial maturity. However, both adaptive and enabling leadership were also correlated with administrative leadership reflecting the entanglement of traditional structures and cultures of health care organizations with bottom-up informal emergent forces. A question that might be asked is: whether administrative leadership maintain the status quo by constraining emergence and self-organization to the detriment of organizational adaptability and learning? On an optimistic note, de Bock et al12 provide a case study of such bottom-up informal complex adaptive forces that successfully shifted clinical decision-making from professional silos into transdisciplinary inter-professional working. These shifts were driven by the internal and external tensions about caring for a longitudinal patient journey beyond technical rescue. The personal power of the nurses who were by the bedside, and their "bottom-up" understanding of the patient's needs, catalysed interdependent interactions and self-organization within the different professional groups. Care was thus adapted to patient-centred approaches beyond reductionist repair modes of thinking. This Forum highlights this emerging implementation of practical, but early stage CAS approaches to improving the outcomes of clinical care and health care more generally. To progress, a vision and practical goals for the shift needed from a conservative medical hierarchical disease focus, toward a more integrated biopsychosocial dynamic interactive ways of thinking about health.3 Tools to enable such implementation are needed, and four different practical approaches to deploy CAS theory in clinical care are highlighted that demonstrate innovation and adaptive thinking. They demonstrate a transition into enabling and adaptive leadership roles from the bottom-up. Yet the paper by Horvat and Filipovic provides some explanation about the slowness of the such transitions related to the challenges to complexity based leadership, with the ever-present dominant conservative health organizations. Administrative leadership models and cultures seeks to maintain the status quo and, for all intense and purposes, stand in the way of innovation and the emergence of "adapting and innovative" processes of care, system organization, and leadership. The International Organization for Standardization, a worldwide federation of national standards bodies (ISO member bodies), states that achievement of sustained success for any organization in a complex, demanding, and ever-changing environment requires enabling and adaptive leadership in health organizations.11 Health care will have to go through a huge cultural change to improve its organizational maturity with enabling and adaptive leadership. There is a need to successfully shape new ways of working and organizing in the evolution of health care. The role of adaptive leadership, as Ron Heifetz pointed out so eloquently, is not to solve problems, but rather to facilitate the necessary adaptive work of the people directly confronting the problems, often in the front-line in health care.13

Complexity science suggests that our current health care delivery system acts as a complex adaptive system (CAS). Such systems represent a dynamic and flexible network of individuals who can coevolve with their ever changing environment. The CAS performance fluctuates and its members' interactions continuously change over time in response to the stress generated by its surrounding environment. This paper will review the challenges of intervening and introducing a planned change into a complex adaptive health care delivery system. We explore the role of the "reflective adaptive process" in developing delivery interventions and suggest different evaluation methodologies to study the impact of such interventions on the performance of the entire system. We finally describe the implementation of a new program, the Aging Brain Care Medical Home as a case study of our proposed evaluation process.

“The study of cas is a difficult, exciting task. The returns are likely to be proportionate to the difficulty.” Holland (2006) On August 9, 2015, cancer took Prof. John Henry Holland away from us. Prof. Holland was a pioneer of Complex Adaptive Systems (CAS) research and a true inspiration. He is known not only for his work on CAS, Holland (1962, 1992)—which he would fondly write as “cas”—but also for his seminal work on adaptation in natural and artificial systems leading to the creation of genetic algorithms and eventually the fields of evolutionary computation, Holland (1995) and Learning Classifier Systems, Holland and Holyoak (1989). Holland was a truly interdisciplinary academic. He had an undergraduate degree in Physics from MIT (1950), an M.A. in Mathematics (1954) and possibly the first ever PhD in Computer Science (1959), both from the University of Michigan—a place where he also subsequently served as a Professor of Psychology, Electrical Engineering and Computer Science. Holland leaves behind his legacy in the form of a large number of thought-provoking articles, video lectures, books, and inspired people—ranging from colleagues, fellows and students to budding complexity enthusiasts. Two of his recent books summarize his views on CAS in both a longer, Holland (2012) as well as a shorter form, Holland (2014). It is easy to foresee that these works will serve not only as a guide to CAS but also guidance for future generations. Holland will indeed be greatly missed. Links to some of his online obituaries are as follows: Melanie Mitchell http://tinyurl.com/qcj22tv National Center for Science Education http://tinyurl.com/pm7ga8y New York Times http://tinyurl.com/p5cd22u Santa Fe Institute http://tinyurl.com/qhkgtxd The Scientist http://tinyurl.com/pns6t64 University of Michigan http://tinyurl.com/obbdpx5 Washington Post http://tinyurl.com/oalp27x

Classical forest management has worked out a series of forest regulation methods with the aim of obtaining the “fully regulated” forest. Considering the forest as a complex biological adaptive system means overcoming the reductionist and mechanist paradigm, and entails a shift towards a systemic approach in silviculture and forest management. The aim of this work is to discuss the objectives and theoretical assumptions of classical forest management methods in the light of the new systemic paradigm. I conclude that managing forests as complex adaptive systems and sustaining their ability to adapt to future changes is possible only if there is also a change in forest management methods so that they are consistent with the new theoretical approach.

PurposeThe aim is to investigate the state of complex adaptive system (CAS) theory in the organizational theory literature and to provide a map for future studies of CAS theory.Design/methodology/approachAbstracts were searched via electronic database and a range of recently published (1996‐2004) books and articles were identified that contained a relatively concise description of CAS. Content analysis is used to deconstruct the CAS descriptions into “component concepts.” Those concepts are analyzed from multiple viewpoints.FindingsThere is no single, shared, sense of CAS theory. Differing understandings of CAS theory are identified based on “expert version” and “most popularly identified concepts.” Also, differences and similarities are identified between an “academic” version of CAS and a version developed by those who are influenced by both academic learning and practical experience.Research limitations/implicationsStudy is limited to concise definitions of CAS, so could be improved by including more lengthy conversations. Additionally, study is limited to organizational theory, so may be less applicable in other disciplines.Practical implicationsWhen working within a CAS framework, academics should specify their CAS perspective to improve clarity of their work. When using a CAS framework to study organizations, researchers should include a comprehensive suite of concepts. Though not described in depth, no effective application of CAS for organizational change were found.Originality/valueFor those who study CAS theory and theory of theory, this paper provides an important benchmark by identifying a bifurcation in the evolution of CAS theory.

Complex adaptive systems (CAS) have been identified as being hard to comprehend, composed of multiple interacting components acting interdependently with overlapping functions aimed at adapting to external/environmental forces. The current theoretical model utilized the natural functions of teams, viewing teams as a complex adaptive system, to develop the structure of the theory of complex adaptive team systems (CATS). The CATS model was formulated around the components of complexity theory (interactions, nonlinearity, interdependency, heterogeneity, complex systems, emergence, self-organizing, and adaptability) to show its utility across multiple domains (the role of leadership, organizational learning, organizational change, collective cognitive structures, innovation, cross-business-unit collaborations). In theorizing the CATS model, a new level of analysis was implemented, the interactions between agents as a move toward emergence in complex systems. The CATS model ultimately provides a model for organizations/institutions to drive knowledge creation and innovation while operating in today's complexity.

Complex adaptive systems are a special kind of system with emergent properties and adaptive capacity in response to external environmental conditions. In this chapter, I investigate the proposition that international environmental law, as a set of multilateral environmental agreements, exhibits the characteristics of a complex adaptive system. This proposition is premised on the scientific understanding that the subject matter displays properties of a complex adaptive system. If so, the legal system may benefit from the insights gained and from being modeled in ways more appropriately aligned with the functioning of the Earth system itself. I provide as context a scientific explanation of the Earth system as a complex adaptive system. I then consider if international environmental law can be understood as a system, which is complex and adaptive. From this exploratory review, I found evidence suggesting that international environmental law is a system with interactive elements. I also found indications of self-organization and emergence, suggesting that international environmental law is a complex system. However, it is still questionable whether the legal system has been autonomously adaptive to and co-evolving with global environmental and geopolitical change in ways that lead to net environmental improvement.

Supporting interoperability in complex adaptive enterprise systems: A domain specific language approach

This paper presents one form of mapping Artificial Immune Systems (AIS) into Learning Classifier Systems (LCS). Artificial Immune Systems can be defined as adaptive systems inspired by theoretical models and principles of the biological immune system and applied to solve problems in the most diverse domains, from biology to computing. Similar to Learning Classifier Systems, already used to model complex adaptive systems, a better understanding of Artificial Immune Systems can be obtained when they are analysed under the perspective of complex adaptive systems. One of the goals here is to determine complementary features of both systems (LCS and AIS), aiming at providing a novel mapping conception. The formal treatment proposed along the paper may then be used to integrate models for complex adaptive systems.

Arecent Institute of Medicine (IOM) report describes a chasm in health care quality that we must cross for patients to receive better care in the 21st century. The report calls for a “systems approach,” drawing on the rapid evolution of knowledge about complex adaptive systems.1 Understanding how complex adaptive systems work can give physicians insights to develop and modify health care systems. By describing a practical application of the ideas about complex adaptive systems to newborn care, we aim to help pediatricians prepare to lead in this field. A complex adaptive system is a collection of individual agents who have the freedom to act, but because the agents are interconnected, action by any agent changes the context for other agents in the system. One familiar example is the buyers in a stock market. In the last century, it was usual to see organizations as mechanical systems: in mechanical systems, if we know what each part of a system does, we can predict perfectly how the whole will respond in a given situation. This is obviously not true of the stock market. A complex adaptive system may display sudden unpredictable shifts in behavior caused by interactions among agents. An essential first step in improving the US health care system is to recognize that its member organizations and individuals, with sublevels nested within and interconnected to each other, make up a complex adaptive system. One of the key attributes of a complex adaptive system is that orderly behavior can emerge among many agents who are acting independently but who share a common drive. For instance, ants, driven to survive, create intricate buildings and foraging systems without any planning by a chief executive ant. So do humans. The citizens of New York City share a drive to eat; with no single individual …

This project investigates the issue of unwarranted escalation of commitment in the toy retailing business. Unwarranted escalation of commitment refers to situations where decision-makers allocate additional resources to failing courses of action. In addition this project aims to identify several complex adaptive system issues within the toy retailing industry in Australia, proposing that the industry satisfies the underlying assumptions of a complex adaptive system. The project is a part of a larger program initiated by Dr. Drew Wollin in pursuing the dual themes of unwarranted escalation of commitment and complex adaptive systems. The Australian toy retailing industry has, since the entrance of the two category killers World 4 Kids and Toys R Us in 1993, been undergoing some major competitive changes. Their entrance was predicted to boost sales and both were aiming for a market share of 20 per cent by late 1995. However, sales remained flat and the objectives were never achieved. After severe price wars and pushed margins, the two giants lost money every year in the period 1993 to 1998, with individual accumulated losses of approximately AU$ 200 million. This project addresses how the organisations persisted as failing ventures. In particular, did they experience unwarranted escalation of commitment to failing courses of action? While there is a lot of contemporary literature in the area of escalated commitment, there is limited research that examines the phenomenon at an organisational level. The most notable exceptions are the case studies by Ross and Staw (1986, 1993) and Newman and Sabherwal (1996). The former investigated a world fair held in Canada (1986) and a decision to set up a nuclear power plant in the US (1993), while the latter examined organisational escalation of commitment in information systems development. This earlier work indicates that models of organisational unwarranted escalation of commitment have not reached theoretical saturation. The two models available, the non-cyclic model (Ross and Staw, 1993) and the cyclic model (Newman and Sabherwal, 1996), are quite different by nature. However, they agree on the escalation determinants. These are project, psychological, social, organisational (structural), and contextual determinants, which are all (to different extents) addressed in this research. Limited amount of theory has investigated complex adaptive systems in organisational and industrial contexts. This study explores whether the theory of complex adaptive systems may contribute to more fully understand the phenomenon of organisational escalation of commitment. A three-level methodological design is proposed in order to address the relevant issues. The macro-research design is analytic induction. This study is part of ongoing research program and performs one iteration of the analytic induction process. The meso-research design is case study research, whereas the micro-research design employed is historical research, based primarily on secondary evidence. Three significant modifications to existing escalation theory are suggested. First, social determinants were found to be important in all phases of the project. As this is one of few studies investigating unwarranted escalation in competitive environments, it is proposed that game theory aspects be incorporated in the social determinants. Second, organisational determinants were significant in the initial commitment decision processes. Third, contextual determinants were to a greater extent present at all stages. This study argues that the Australian toy retailing industry is a complex adaptive system, and displays behaviour expected from such systems. It is suggested that more research be done aiming to provide explanation of organisational unwarranted escalation in terms of behaviour expected from complex adaptive systems.

Bioeconomy is a complex economic system, the functioning of which leads to the emergence of a synergy effect due to the effective combination of natural resources and technologies along with the market-based, social, and political components. In a bioeconomic system, established links are formed between traditional industries based on the use of fossil natural resources and modern ones that have not previously intersected. As a result, one industry uses by-products of another industry as raw materials, thus forming a closed production cycle. The purpose of the article is to develop an author’s vision of the conceptual foundations to substantiate the sustainability of the bioeconomic system, taking into account its specific characteristics, modern challenges and potential of the country. The article reveals the features of transformations of socioeconomic systems into a complex adaptive bioeconomic system in a dynamic perspective. The complexity and complicacy of transformation management lies in the relationship and interaction of the elements of the system, as well as between the very system and its environment. A theoretical analysis of the evolution of the concept of «bioeconomy» is carried out and its basic properties are determined. The interpretation of the bioeconomy as a complex adaptive system that ensures the achievement of sustainable development goals is substantiated. Based on empirical data from the State Statistics Service of Ukraine and the European Commission, potential sectors of the bioeconomic sector of Ukraine have been identified, and the dynamics of the contribution of potential sectors of Ukraine’s economy to the national economy has been analyzed according to the main indicators: generation of value added and creation of jobs. Based on the theoretical and analytical research, it is proved that bioeconomy, as a complex system combining several sectors of the economy, can adapt to changes occurring in the environment.

Complex Adaptive Socio-Technical Systems The Role of Socio-Technical Networks in New Product Development

Complexity science is a new field of science that help us to understand complex social phenomenon such as economies, traffics, wars, epidemics, mass actions, etc. On the other hand, multi-agents simulation (MAS) methods are useful and powerful tools for complex system science. They come to be applied to studies and analysis of complex systems along with the development of computers. It can be said that the perspectives of complex systems and the skills for MAS are very useful knowledge for students who should live and work in the complex modern society. Then the author tried to design and conduct an education program dealing with complex system science and MAS as liberal arts targeting at students from all faculties. It's a project-based learning program in which students investigated and discussed What is complex systems? and develop MAS. In general, it is difficult for students who haven't gained the any special knowledge of computing to build MAS. Then the author applied the generic multi-agent simulation platform to this program. With artisoc, students can build MAS easily. In this paper, the practical example of PBL program with artisoc and its result are introduced.