Gå til hovedinnhold
0
Hopp til hovedinnhold

Social simulation: new ways of analyzing a society

Ivan Puga Gonzalez started out using agent-based computer models in order to understand the social behavior of primates. Now he applies such models to analyze and understand human social behavior. Can a computer model tell us anything about human social phenomena?

Artikkelen er mer enn to år gammel, og kan inneholde utdatert informasjon.

Postdoctoral researcher Ivan Puga Gonzalez uses computer models and simulation in order to analyze social behavior and interaction.

Social simulation will be the focus of an upcoming conference 30 September, arranged by the Social Simulation research group at UiA and co-sponsored by Center for Strategic Studies at Tufts University and the MIT Energy Initiative.  The goal of the event called "Climate and Conflict: Seeking Solutions through Social Simulation" is to bring together experts on climate and conflict from around the world to help develop computer models and simulations that will provide policy-relevant insights for both governments and NGOs. 

Postdoctoral researcher Ivan Puga Gonzalez is convinced that social simulation can be useful to both researchers and policy makers. He thinks it’s important to show how modelling can be used as a virtual laboratory to better understand human social phenomena and try out the impact of new policies before their implementation in the real-world.

Mimicking societies

Puga Gonzalez works at the Department of Global Development and Planning and is involved in several research projects such as the spread of religious beliefs, integration of minorities, and bullying behavior. His approach to these questions is through agent-based models and social simulation. This means that he’s mimicking an actual society in a computer program, a society that he can then run experiments upon. All these projects are based on current psychological theories of human social behavior. In one of them, he uses a computer model that relies on so-called CRED-theory. This theory is one of the leading social psychological theories in the scientific study of secularization and religion, explaining how people adopt or abandon certain worldviews.

The model constitutes an artificial society where the agents interact and behave based on a cognitive architecture informed by CRED theory. The model is fed by large amounts of statistical data and information about societies’ social structural and human behavior. The agents in it are individuals that attend school, work, marry and reproduce, and have variables related to their worldview, age, education, employment and more. In this model he can then explore questions such as: what are the most important factors leading to societies where all individuals hold a secular or religious worldview?

Agent-based models are computer programs that can represent societies, industries, ecosystems etc. Usually, they have a user interface where researchers can visualize the simulation and change its conditions. This allows researcher to experiment with hundreds or even thousands of different scenarios.

A continuous conceptualization and validation process

- Social systems are inherently complex, its behavior depends on the interactions among many interrelated processes which makes it difficult to understand how a change in one process may impact others and the whole system. By putting these processes in a controlled environment, such as a computer model, and growing the phenomena from the micro-level (individuals’ interactions), it is easier to understand how and why a phenomenon at the macro-level (society) emerges, Puga Gonzalez explains.

When building a model, a big part of the job is a cycle between model conceptualization and validation. These stages rely on the involvement of the expertise of others and the use of empirical data. “Subject matter experts” – researchers with specific knowledge about the relevant matter – play a big part during conceptualization, and involving these is key since they inform about the social processes and human behavior that are relevant to include in the model. Once conceptualized, the model is coded and run, and its output goes through a reality check - model validation. If the output of the model is in contrast with reality, the conceptualization is revised, and the cycle starts again. Only when the output is a good approximation of reality, one may start running experiments with the model. By experimenting with the model (e.g. adding or omitting specific processes or interactions), researchers may then find ways that lead the whole system to a desired scenario.

Groundbreaking in social sciences

Although the use of modeling and artificial societies has increased rapidly within the social sciences during the recent years, the phenomenon is still met by skepticism by many social scientists. The degree of complexity that these models can comprise, is often underestimated by the sceptics, according to Puga González. – Researchers without any experience with modelling tend to think that the complexity of the issues and processes that they deal with in social science is impossible to capture in a virtual world. To a certain degree I can agree with that, but the point is that in a model you can capture the parts that are essential and look at them separately.

He stresses that modelling has its weaknesses as any other methodology, but that the technology is a tool for social scientists and policy professionals to explore and test their hypotheses in a systematical way. - I will argue that one of the unique values is being able to experiment with factors in a way that you can’t do in real life. Both the conceptualization and experimenting stages can be useful for discovering flaws or gaps in one’s own theory or assumptions.

UiA conference on climate and conflict

The research field of social simulation is getting increased attention in Europe, and Puga Gonzalez and his colleague professor F. Leron Shults will soon be welcoming several international research partners to The Climate and Conflict consortium 28-29 of September in Kristiansand. In the public presentation of this consortium 30 September, they will address how artificial intelligence can prevent conflict and promote peaceful coexistence in the face of climate change.

- These challenges serve as good examples of how social simulation can be used, Puga Gonzalez explains. Through models we can first reproduce problems related to climate change such as migration, armed conflicts, changes in land use, water scarcity, etc., and then we can experiment with interventions in order to find potential solutions that may prevent or mitigate the likelihood of such events in the real world. 

For details on the public presentation 30 September, check out the event on Facebook.