The use of modeling and simulation (M&S) methodologies is growing rapidly across the psychological and social sciences. After a brief introduction to the relevance of computational methods for research on human cognition and culture, we describe the sense in which computer models and simulations can be understood, respectively, as “theories” and “predictions.” Most readers of JoCC are interested in integrating micro- and macro-level theories and in pursuing empirical research that informs scientific predictions, and we argue that M&S provides a powerful new set of tools for pursuing these interests. We also point out the way in which M&S can help scholars of cognition and culture address four key desiderata for social scientific research related to the themes of clarity, falsifiability, dynamicity, and complexity. Finally, we provide an introduction to the other papers that comprise this special issue, which includes contributions on topics such as the role of M&S in interdisciplinary debates, shamanism, early Christian ritual practices, the emergence of the Axial age, and the social scientific appropriation of algorithms from massively multiplayer online games.
1 The Computational Study of Cognition and Culture
This special issue aims to demonstrate the value of computer modeling and simulation (M&S) for studying human cognition and culture. By offering several examples of computational models and social simulations, and discussing the challenges and opportunities associated with their construction and execution, we hope to provide readers with a better understanding of the way in which these relatively new methodological tools can help to explain the mechanisms at work in religious minds and groups. M&S tools have played an important role in the natural sciences for over half a century, and have been extensively used for business and military purposes (Tolk, 2012). In fact, these tools have had such a revolutionary effect on a wide variety of scientific disciplines (Humphreys, 2006) that they have been called the “third pillar” of science, alongside theory and experimentation (Benioff & Lazowska, 2005).
In the last couple of decades, the use of M&S methodologies has expanded rapidly in the social sciences as well (Alvarez, 2016; Gilbert & Troitzsch, 2005; Squazzoni, 2012). Even more recently simulation tools have been making inroads in a variety of disciplines in the humanities, including history and philosophy (DeLanda, 2011; Elsenbroich & Gilbert, 2014; Wildman, Fishwick, & Shults, 2017; Youngman & Hadzikadic, 2014). As the articles in this special issue show, computational approaches bring their own unique set of challenges, but they also provide new opportunities for clarifying and integrating scientific theories about human cognition and culture.
Scholars with special interest in the cognitive dynamics that shape human experience may find themselves attracted to the discussion of cognitive architectures in this literature. Many of the early and most influential of these architectures focused primarily on the mechanisms driving representation, deliberation and learning (e.g., Sun, 2006). More recent efforts in the field have often attempted to incorporate the affective dimensions of human cognition into their architectures (Gratch & Marsella, 2014; Scherer, Bänziger, & Roesch, 2010). Scholars whose research tends to be oriented more toward the cultural aspects of human experience are more likely to be interested in the way in which these powerful tools for simulating social experiments have increasingly been applied to phenomena such as the differentiation and transmission of norms and cultures (Dignum and Dignum 2014).
Many readers of Journal of Cognition and Culture, however, are interested in both cognition and culture – and in the way in which their reciprocal interaction is shaped by individual differences across diverse social contexts. Scholars who are willing to attempt multi-disciplinary explanations of social phenomena are faced with the challenge of making sense of the relationship between micro-level (cognitive) mechanisms and variables and macro-level (cultural) dynamics and patterns. Several of the articles in this special issue provide illustrations of the distinctive way in which M&S can help to connect these two levels. System-dynamics models (SDMs), as one might expect from the name, often focus primarily on the macro-level (structural) dynamics of systems, such as those systems that constitute and regulate human societies. Agent-based models (ABMs), on the other hand, are constructed by focusing initially at the micro-level (agent) dynamics of systems, such as the characteristics and behavioral interactions of the entities in the artificial society. Such entities may include “agents” at the sub-individual level (e.g., cognitive mechanisms), at the individual level (e.g., humans), or at a super-individual level (e.g., social groups). Both SDMs and ABMs can, in principle, allow for researchers to experiment on the effects of environmental variables through the use of markers of resource availability or alterable parameters related to carrying capacity (for example). Even meso-level factors can be explored by incorporating networks into agent-based models.
Social simulation provides a unique way to study the emergence of the complex adaptive systems involved in human cognition and culture. Here we are not dealing with “spooky emergence” but with the capacity of computer models to generate the phenomena of interest by having complex agents act (and interact) by means of multiple pre-defined rules. As Joshua Epstein put it in his discussion of the role of agent-based models in a generative social science: “If you didn’t grow it, you didn’t explain its emergence” (Epstein, 2006, p. 8). Generative computational models aim to explain complex social phenomena by “growing” macro-level phenomena from “bottom up.” Emergence in this sort of context refers to the way in which stable macroscopic patterns or collective structures can arise from the local interaction of individual agents at the micro-level (Epstein & Axtell, 1996).
The term “emergence” is utilized in different ways in the humanities, social sciences and fields such as engineering, and it is important to be aware of the semantic shifts when reaching across disciplines (Wildman and Shults 2018). The main point here, however, is that such methodologies can shed light on the conditions under which – and the mechanisms by which – individual cognitive processes can generate (or “evoke”; Tooby & Cosmides, 1992) collective cultural patterns.
2 Models as Theories, Simulations as Predictions
The usefulness of M&S may not be readily apparent to many in the social sciences. However, scholars of cognition and culture are already adept at constructing and criticizing theories and are accustomed to searching for variables or factors that best predict the psychological or sociological phenomena of interest. Computer modeling and simulation tools provide a new way of enhancing both activities. From an information processing perspective, we can view models as the implementation of theories (or integrated sets of theories) in a computational architecture and simulations as predictions (outcome data) generated by multiple runs of that architecture under various parameter settings.
Computer models are re-presentations or translations of a theory (or theories) into a computer language. In the social sciences and humanities, theories are typically presented in narrative form. And such narratives often require translation. For example, Durkheim’s theory of religion must be translated into English for use by scholars in the English-speaking world. Such theories (or components of them) can also be translated into a computer programming language (e.g., Java or C++), which then allows them to be explored through simulation experiments. This translation from narrative to computational architecture requires that these theories be clarified in such a way that their component parts and the relations among them are susceptible to formalization and coding. This process of “translating” a theory into a computer programming language often forces scholars to new levels of conceptual clarity. They must be exceptionally precise in their explanation of the key variables and their hypothesized interactions. This sort of translation sometimes requires simplifications of a theory or abstractions that capture only some of its most relevant aspects. Once a theory is adequately implemented within a computer model, however, the latter can be taken as a representation of the former.
What about prediction? Once the computational architecture of a model has been calibrated and verified to be an acceptable representation of the theory, the output of its simulation experiments can be understood as predictions. In other words, the simulation data can be explored to discover the mechanisms and interactions in the model that lead to various outcomes in an artificial society. Such predictions can then be compared to observations of the target phenomenon in the real world, a process called “validating” the model. Validation is a critical step that can involve direct collaboration between scholars with expertise in other fields such as psychology, archaeology, history, or anthropology. Using a variety of methods from these fields, including (but not limited to) experimentation, database analyses, fieldwork, and network and big-data analyses, we can compare the output of the model to observations in the real world to determine whether it is a useful representation of the latter. In short, verification is the process by which a model is determined to be an appropriate implementation of a theory. Verification is meant to ensure that the model addresses the appropriate components of a theory and that its critical elements (such as the causal effects specified by the theory) are appropriately implemented. Validation is the process by which a model is determined to be a correct representation of the target phenomenon. Validation is meant to ensure that the model addresses the appropriate range of observations of a target phenomenon and that the dynamics noted in real-world observations are isomorphic to the output of the model.
3 Desiderata for Social Scientific Hypotheses and Experimentation
Developments in computer modeling and simulation can also help scholars of cognition and culture address at least four of the criteria against which scientific theories and experimental designs are often measured. These desiderata are related to the themes of clarity, falsifiability, dynamicity, and complexity.
First, constructing a computer model to test one’s scientific hypotheses requires a high level of conceptual clarity. Terms are often polysemous across and even within disciplines, perhaps even more so in the social sciences and humanities than in the natural sciences. Computational modeling forces precision in the definition of variables and the formulation of the algorithms that drive agent behaviors or system dynamics. Moreover, computer programming languages are intolerant of logical inconsistencies, which means they are likely to flag errors if (for example) a variable is treated in different ways in various parts of the model. The process of creating a computer model itself forces the subject matter expert to clarify terms and relations that he or she might otherwise leave undefined or underspecified.
A second desideratum is the falsifiability of hypotheses. In the social sciences theoretical claims are all too often presented in ways that do not make it clear how they could be tested and potentially falsified. This is problematic because if a theory cannot be proven wrong, it cannot be accepted as useful for explaining (as opposed to interpreting) the social phenomenon in question. Furthermore, theories that are not susceptible to falsification can easily serve as scaffolding for ideologies which, while they may be hermeneutically helpful in some domains, can sometimes hinder researchers from combatting their implicit biases. Once one validates the isomorphism of a theory and its implementation within a computer model, the latter permits hypothetico-deductive reasoning about the former. We can formulate hypotheses about the relationship between the causal architecture that comprises the theory and expected observations in the simulation results. Since computer experiments can be run by other researchers under the same conditions, the falsification of claims can be replicated, and other predictions can be explored across (computational) laboratories. Furthermore, the falsification of the model (theory) – or at least some part of it – occurs when the output of the model is not consistent with the hypotheses. This shifts the role of subject matter experts in the social sciences and humanities from one in which they merely provide data to be included within a model to one in which they can help guide the decision to accept or reject a model (or aspects of the model) during the verification and validation processes. This approach more directly engages scholars from the social sciences and humanities in the scientific process, enabling them to participate in the confirmation or falsification of theories as members of robustly interdisciplinary research teams.
Third, social scientific theories can be judged by the extent to which they are able to make sense of the dynamic relations among the variables they are meant to explain. Most statistical methods provide analyses of variables in a particular state, that is, of the correlation of static variables frozen in time and space. Computer models, on the other hand, can allow for the exploration of the way in which time itself plays a role in the hypothesized movements of simulated agents through space. If statistics are pictures, then simulations are movies. And if a picture is worth a thousand words, how much more valuable to a scientist is a tool that permits experimentation through the “cinematic” production and editing of the “stories” of dynamic artificial societies? While the temporal dimension is all too often neglected or overly simplified in most statistical analyses, the time dynamics of social phenomena are explicitly incorporated into computational models. Simulation experiments formalize and clarify the role played by the passage of time in the interactions between variables hypothesized by a theory.
The final desideratum we will mention here is related to complexity. It is often the case that variables treated in the social sciences have non-linear relationships; for example, variables may be wrapped up within feedback loops, be characterized by iterative and/or recursive processes and have multiple causes or effects at different levels. Computer models are able to address and take into account all of these sorts of complexity in a way that is not possible with other methods. This is particularly important when dealing with the causal relationships among multiple cognitive and cultural variables within the complex systems that make up human life. For example, it is commonly accepted that evolved cognitive mechanisms shaped by selective pressures from the environment interact in complex ways to produce (or evoke) macroscopic cultural patterns. The latter, in turn, alter the social (and natural) environment and can affect the operation or expression of cognitive mechanisms. These kinds of complex feedback loops between cognitive mechanisms, individuals, and their environment can sometimes border on the chaotic, but they can be studied through the use of agent-based models that incorporate the information input-output streams from the relevant cognitive and environmental systems (Lane, 2013, 2017a).
Computer modeling and simulation tools enable us to express our hypotheses about the complexities of cognitive and cultural systems with greater clarity, to analyze them in ways that better account for their inherently temporal dynamics, and to render them falsifiable through simulation experiments. All of this presupposes collaboration, rather than competition, with other methodologies relevant to the study of cognition and culture.
4 Toward a New Computational Social Science of “Religion”
All the articles in this special issue more or less explicitly address religion, a topic in which many readers of the Journal of Cognition and Culture have long been interested. By applying the tools of M&S to religious phenomena, these papers contribute to a growing body of literature in what we might call the computational science of religion (and non-religion; Shults, Gore, Lemos, & Wildman, 2018). Early efforts in the computational science of religion included models of new religious movements (Upal 2005), artificial intelligence models of religious cognition (Bainbridge, 2006), simulations of the persistence of religious regionalism (Iannaccone & Makowsky, 2007), and agent-based models of groups with costly signals (Wildman & Sosis, 2011). In the last few years, scholars have increasingly pointed to the usefulness of these sorts of tools for the scientific study of religion (Braxton, Upal, & Nielbo, 2012; Lane, 2013). Outside of this special issue, the research on modeling and simulation has continued to grow at a very fast pace in recent years addressing themes ranging from supernatural punishment (Lane 2017b), mutually escalating religious violence (Shults et al., 2017), the relationship between education, existential security and religiosity (Gore, Lemos, Shults, & Wildman, 2018), the dynamics of religious radicalization (Shults, 2018; Shults & Gore, 2018), and the effects of ritual on identity and social stability (Lane 2018).
In the first article of this special issue, Andreas Tolk and colleagues propose “human simulation” as a scientific lingua franca that can help draw scholars in the humanities and social sciences into dialogue with scholars from other fields who use computational methods. Human simulation can refer to the simulation of humans, the design of simulations for human use, or simulations in which humans can participate as agents within a virtual society. They explore the potential of this approach for integrating the efforts of the relevant research communities, including humanities scholars in traditional religious studies programs. The article concludes with a discussion of some of the computational, epistemological and hermeneutical challenges involved in this endeavor.
The second article sets out a computational model of the “Axial Age,” the period from approximately 800–200 BCE during which significant changes in civilizational forms occurred in west, south, and east Asia. Here Shults and colleagues present a conceptual architecture that reconstructs and integrates three distinct hypothesized pathways meant to explain the emergence of axial civilizations, each reconstructed on the basis of theoretical developments and empirical findings in a wide range of disciplines. This theoretical integration is implemented within a system-dynamics model, which is able to simulate the pathway through which the majority of a population converts to an axial worldview and to show the conditions under which that transition does or does not occur.
Third, Vojtech Kaše and his coauthors from the GEHIR research group demonstrate the relative advantages of agent-based and system-dynamics models for modeling the cultural transmission of religious rituals. They focus on social function and cognitive attraction as two of the most important driving factors that independently affect the process of ritual transmission. Engaging with the growing body of literature in cognitive historiography and the use of historical data in contemporary investigations of religion and culture (e.g., The Database of Religious History 2015), they focus their investigation on the selection of early Christian ritual meals. They find that cognitive attraction has a greater force on the selection of ritual behaviors than social function.
In the fourth article, Connor Wood and Saikou Diallo present an agent-based model that enables the exploration of factors that could underpin differences between societies that produce female-dominated spirit possession cults, versus those that produce male-dominated shamanism. Because of proposed gender differences in how social and psychological pressures effect individuals they hypothesize that different types of hierarchical arrangements affect the gender balances of shamanistic practitioners in societies. Under plausible parameter configurations the model is able to simulate the emergence of patterns noted in the shamanism literature.
William Bainbridge, one of the most well-known pioneers and founders of a computational science of religion, provides the final article in this collection. He discusses examples of the simulation of religion within massively multiplayer online games, such as World of Warcraft, and argues that conceptually and algorithmically creative approaches could be taken from the commercial world and used in scientific research. The results of his research into these gaming worlds (Bainbridge 2012, 2007) and the complex architectures underlying the simulated characters within these worlds, demonstrate the capacity of computational methods for expressing the richness and complexity of religiosity.
The computational approach to religion appears to be in a toddler phase when compared to the use of computer modeling and simulation in the natural sciences and engineering. However, the community of scholars interested in applying these methods to research on religion is lively and growing rapidly. This special issue demonstrates what can happen when cognitive and social scientists are fortunate enough to collaborate with some of the world’s leading experts in M&S, including experts based in the corporate world (Tolk, The MITRE Corporation) and the government (Bainbridge, NSF). All too often, such opportunities for collaboration are neglected – or even discouraged. We hope that the studies showcased here will not only shed light on old problems within our own fields, but also lead to innovations that can be used in other disciplines, as well as by policy professionals interested in using M&S to help address some of the daunting societal challenges we all face in the contemporary world.
Bainbridge, William Sims. 2006. God from the Machine: Artificial Intelligence Models of Religious Cognition. Lanham, MD: AltaMira Press.
Bainbridge, William Sims. 2007. “The Scientific Research Potential of Virtual Worlds.” Science 317 (5837): 472–77. https://doi.org/10.1126/science.1146930.
Benioff, Marc R., and Edward D. Lazowska. 2005. “Computational Science: Ensuring America’s Competitiveness.” National Coordination Office for Information Technology Research & Development. https://www.nitrd.gov/pitac/reports/20050609_computational/computational.pdf.
- Search Google Scholar
- Export Citation
. , and Benioff, Marc R. Edward D. Lazowska . “ 2005 Computational Science: Ensuring America’s Competitiveness.” National Coordination Office for Information Technology Research & Development. https://www.nitrd.gov/pitac/reports/20050609_computational/computational.pdf.
Braxton, Donald M., M. Afzal Upal, and Kristoffer L. Nielbo. 2012. “Computing Religion: A New Tool in the Multilevel Analysis of Religion.” Method & Theory in the Study of Religion 24 (3): 267–90. https://doi.org/10.1163/157006812X635709.
Dignum, and Dignum. 2014. Perspectives on Culture and Agent-Based Simulations: Integrating Cultures. Vol. 3. Studies in the Philosophy of Sociality. Cham: Springer International Publishing.
Epstein, Joshua M., ed. 2006. Generative Social Science: Studies in Agent-Based Computational Modeling (Princeton Studies in Complexity). Princeton, NJ: Princeton University Press.
Epstein, Joshua M., and Robert L. Axtell. 1996. Growing Artificial Societies: Social Science from the Bottom Up. Washington, D.C.: Brookings Institution Press.
Gore, Ross, Carlos Lemos, F. LeRon Shults, and Wesley J. Wildman. 2018. “Forecasting Changes in Religiosity and Existential Security with an Agent-Based Model.” Journal of Artificial Societies and Social Simulation 21 (1): 1–31. https://doi.org/10.18564/jasss.3596.
- Search Google Scholar
- Export Citation
. , Gore, Ross , Carlos Lemos , and F. LeRon Shults Wesley J. Wildman . “ 2018 Forecasting Changes in Religiosity and Existential Security with an Agent-Based Model.” Journal of Artificial Societies and Social Simulation 21( 1): 1– 31. https://doi.org/10.18564/jasss.3596.
Gratch, Jonathan, and Stacy Marsella. 2014. Social Emotions in Nature and Artifact. Oxford Series on Cognitive Models and Architectures. Oxford University Press.
Iannaccone, Laurence R., and Michael D. Makowsky. 2007. “Accidental Atheists? Agent‐Based Explanations for the Persistence of Religious Regionalism.” Journal for the Scientific Study of Religion 46 (1): 1–16. https://doi.org/10.1111/j.1468-5906.2007.00337.x.
- Search Google Scholar
- Export Citation
. , and Iannaccone, Laurence R. Michael D. Makowsky . “ 2007 Accidental Atheists? Agent‐Based Explanations for the Persistence of Religious Regionalism.” Journal for the Scientific Study of Religion 46( 1): 1– 16. https://doi.org/10.1111/j.1468-5906.2007.00337.x.
Lane, Justin E. 2013. “Method, Theory, and Multi-Agent Artificial Intelligence: Creating Computer Models of Complex Social Interaction.” Journal for the Cognitive Science of Religion 1 (2): 161–80.
Lane, Justin E. 2017a. “Can We Predict Religious Extremism?” Religion, Brain & Behavior 7 (4): 299–304. https://doi.org/10.1080/2153599X.2016.1249923.
Lane, Justin E. 2017b. “Strengthening the Supernatural Punishment Hypothesis through Computer Modeling.” Religion, Brain & Behavior 8 (3): 12–22. https://doi.org/10.1080/2153599X.2017.1302977.
Lane, Justin E. 2018. “The Emergence of Social Schemas and Lossy Conceptual Information Networks: How Information Transmission Can Lead to the Apparent ‘Emergence’ of Culture.” In Emergent Behavior in Complex Systems Engineering: A Modeling and Simulation Approach, edited by Saurabh Mittal, Saikou Y. Diallo, and Andreas Tolk, 1st Edition, 329–256. New York: John Wiley & Sons, Inc.
- Search Google Scholar
- Export Citation
Lane, Justin E. . “ 2018 The Emergence of Social Schemas and Lossy Conceptual Information Networks: How Information Transmission Can Lead to the Apparent ‘Emergence’ of Culture.” In Emergent Behavior in Complex Systems Engineering: A Modeling and Simulation Approach, edited by , , Saurabh Mittal , and Saikou Y. Diallo Andreas Tolk 1st Edition, 329– 256. New York: John Wiley & Sons, Inc.
Scherer, Klaus R., Tanja Bänziger, and Etienne Roesch. 2010. A Blueprint for Affective Computing: A Sourcebook and Manual. Oxford University Press.
Shults, F. LeRon. 2018. “Can We Predict and Prevent Religious Radicalization?” In Processes of Violent Extremism in the 21st Century: International and Interdisciplinary Perspectives, edited by Gwenyth Øverland. Cambridge: Cambridge Scholars Press.
Shults, F. LeRon, and Ross Gore. 2018. “Modeling Radicalization and Violent Extremism.” Proceedings of the 2018 Social Simulation Conference, 1–4.
Shults, F. LeRon, Ross Gore, Carlos Lemos, and Wesley J. Wildman. 2018. “Why Do the Godless Prosper? Modeling the Cognitive and Coalitional Mechanisms That Promote Atheism.” Psychology of Religion and Spirituality.
Shults, F. LeRon, Ross Gore, Wesley J. Wildman, Justin E. Lane, Chris Lynch, and Monica Toft. 2017. “Mutually Escalating Religious Violence: A Generative and Predictive Computational Model.” Proceedings of the 2017 Social Simulation Conference.
Sun, Ron. 2006. “The CLARION Cognitive Architecture: Extending Cognitive Modeling to Social Simulation.” In Cognition and Multi-Agent Interaction: From Cognitive Modeling to Social Simulation, edited by Ron Sun, 79–102. Cambridge: Cambridge University Press.
- Search Google Scholar
- Export Citation
. Sun, Ron . “ 2006 The CLARION Cognitive Architecture: Extending Cognitive Modeling to Social Simulation.” In Cognition and Multi-Agent Interaction: From Cognitive Modeling to Social Simulation, edited by , Ron Sun 79– 102. Cambridge: Cambridge University Press.
The Database of Religious History. 2015. “The Database of Religious History.” The University of British Columbia [Website]. 2015. http://www.religiondatabase.arts.ubc.ca/.
Tooby, John, and Leda Cosmides. 1992. “The Psychological Foundations of Culture.” In The Adapted Mind: Evolutionary Psychology and the Generation of Culture, edited by Jerome H. Barkow, Leda Cosmides, and John Tooby, 19–136. Oxford & New York: Oxford University Press.
- Search Google Scholar
- Export Citation
. , and Tooby, John Leda Cosmides . “ 1992 The Psychological Foundations of Culture.” In The Adapted Mind: Evolutionary Psychology and the Generation of Culture, edited by , , Jerome H. Barkow , and Leda Cosmides John Tooby 19– 136. Oxford & New York: Oxford University Press.
Upal, M. Afzal. 2005. “Simulating the Emergence of New Religious Movements.” Journal of Artificial Societies and Social Simulation 8 (1).
Wildman, Wesley J., Paul A. Fishwick, and F. LeRon Shults. 2017. “Teaching at the Intersection of Simulation and the Humanities.” Edited by W. K. V. Chan. Proceedings of the 2017 Winter Simulation Conference, 1–13.
Wildman, Wesley J., and F. LeRon Shults. 2018. “Emergence: What Does It Mean and How Is It Relevant to Computer Engineering?” In Emergent Behavior in Complex Systems Engineering: A Modeling and Simulation Approach, edited by Saurabh Mittal, Saikou Diallo, and Andreas Tolk, 21–34. Hoboken, NJ: John Wiley & Sons.
- Search Google Scholar
- Export Citation
. , and Wildman, Wesley J. F. LeRon Shults . “ 2018 Emergence: What Does It Mean and How Is It Relevant to Computer Engineering?” In Emergent Behavior in Complex Systems Engineering: A Modeling and Simulation Approach, edited by , , Saurabh Mittal , and Saikou Diallo Andreas Tolk 21– 34. Hoboken, NJ: John Wiley & Sons.
Wildman, Wesley J., and Richard Sosis. 2011. “Stability of Groups with Costly Beliefs and Practices.” Journal of Artificial Societies and Social Simulation 14 (3).
Youngman, Paul A., and Mirsad Hadzikadic. 2014. Complexity and the Human Experience: Modeling Complexity in the Humanities and Social Sciences. Pan Stanford.