This research project is funded by The Research Council of Norway, and will run from 1 September 2020 through 28 February 2022. The overarching goal is to assist in the development of strategies to combat the spread of misinformation, stigma, and fear in the population in the wake of pandemic crises such as COVID-19.
The outputs of the project will include user-friendly artificial intelligence tools that will enable municipalities and other governmental agencies (1) to analyze and forecast the societal effects of their public health responses and social countermeasures to pandemics, and (2) to experiment with alternative intervention strategies for “flattening the curve” of psychologically and politically debilitating social contagion within an empirically validated “digital twin” (artificial society) before trying them out in the real world.
EmotiCon will utilize survey and other data to analyze the role of social media in fostering anxiety and explore the cultural dimensions of the societal reaction to the disease by running simulation experiments that can reveal the mechanisms and forecast the patterns in which individuals and communities (including vulnerable groups such as the elderly) understand and react to COVID-19. Once calibrated using quantitative and qualitative data, and validated in dialogue with the municipality reference group and other subject matter experts, the multi-agent artificial intelligence EmotiCon model will also inform stakeholder discussions and decisions in relation to public health and risk communication in reaction to future pandemic threats.
Our research teams have already published computational models (here and here) that can simulate the effect of disease and social contagion threats on the attitudes and behaviors of human populations and produce emergent society level patterns validated in relation to data from the Christchurch earthquake in New Zealand and the escalation of conflict in Northern Ireland and India. The agents in those simulations have cognitive architectures and weighted social network ties that model the effects of well tested social psychological theories such as “terror management theory” and “social identity theory” that are relevant for understanding responses to natural disasters such as pandemics. Our plan is to adapt these models to simulate the emotional and behavioral effects of disease contagion threats under a wide variety of parameters relevant to our user reference group (municipalities in Norway). These new tools will enable stakeholders to identify and avoid thresholds that can lead to tipping points at which a population’s reaction to governmental messages and social restrictions spills over in psychologically or economically problematic ways.
The principal investigator of the project is F. LeRon Shults, research professor at the Center for Modeling Social Systems (CMSS) at NORCE and professor at the Institute for Global Development and Social Planning at the University of Agder in Kristiansand, Norway. For a fuller description of the work packages, see below. The leader of the “social media analysis” work package is Justin E. Lane, CTO of Prospectus Solutions. The leader of the “Norwegian survey analysis” work package is Roger Normann, director of CMSS and chief scientist at NORCE. The leader of the work package on “agent-based modeling” is Patrycja Antosz, senior research at CMSS and a post-doctoral fellow at the University of Agder. The core team will be guided throughout the process by a user reference group composed of health policy professionals associated with cities throughout Norway.
Step 1: Social Media Analysis
During the first phase of the EmotiCon project, we will download and analyze millions of networked social media messages related to the COVID-19 crisis. These messages will be coded with variables related to misinformation, stigma, and anxiety so that we can explore the ways in which – and the conditions under which – they spread in the wake of the pandemic.
Step 2. Norwegian Survey Analysis
The second phase of the EmotiCon project will involve the construction and execution of surveys that will produce panel data related to the attitudes and behaviors of Norwegians in reaction to governmental communications and policies dealing with COVID-19. The development of the survey will be informed by the social media analysis described above, and inform the construction of the agent-based model described below.
Step 3. Agent-Based Modeling
The final phase of the EmotiCon project is the development of a multi-agent artificial intelligence model and execution of policy-oriented simulation experiments that can be used by members of our municipality user reference group to identify and avoid the thresholds and tipping points that can exacerbate the negative psychological, political, and economic effects of social contagion when a population is under threat of disease contagion. Rising to the challenge will require new knowledge about the relevant causal factors within the complex systems that constitute and regulate emotional and behavioral contagion as well as new experimental methods that can incorporate a broader scope of multi-disciplinary conceptual insights and provide a greater capacity for computational analysis and synthesis. This is the ultimate goal of our EmotiCon agent-based model.