National Institute of
General Medical Sciences

November 6-7, 2008

At its May 2007 meeting, the National Advisory General Medical Sciences Council advised the institute to explore the science of modeling social behavior as a potential new program. Social dynamics influence many biological processes of interest to NIGMS, including the organization of genetic information, transmission of disease, development of pharmaceuticals, and expanding the diversity of the scientific workforce. NIGMS is particularly well situated to consider support for mathematical and computational modeling of social behavior.

On November 6-7, 2008, NIGMS held an informational conference to identify obstacles and opportunities in the emerging field of modeling social behavior. The organizing committee selected several specific topics as exemplars of modeling in the social sciences. The purpose of the meeting was explore these areas of research, not in the details of content, but as examples of how concepts emerge, what questions are central, the nature of evidence, how analytical methods and modeling are used, and how the field connects to other disciplines.

On the first day of the meeting, speakers were asked to address the following questions:

• How do scientists conceptualize this field?
• What are the central scientific questions?
• What kind of evidence is collected (e.g., biochemical data, behavioral observations, answers to surveys, geospatial information)?
• What are the key findings so far?
• What analytic methods, especially modeling, are used to develop deeper understanding of the biological concepts?
• What disciplines contribute to the field?

The second day consisted of more in-depth discussion of two topics, prejudice and resilience, with the goal of discovering how various modeling and experimental approaches can contribute to understanding these social phenomena.

The NIH Office of Behavioral and Social Science Research supported the videotaping of the meeting (http://videocast.nih.gov).

Meeting Summary

Speakers agreed that modeling is a powerful tool for studying social processes relevant to NIH's mission, in large part because modeling forces us to make our assumptions explicit. As Josh Epstein pointed out in his introduction, "Anyone who ventures a projection or imagines how a social dynamic would unfold is running some model." Often our models are implicit, based on unstated assumptions, unknown relationships to data, and unverified consequences. Good modeling grows from a close association with empirical data, in which each informs the other.

People often assume that the goal of modeling is prediction. While it may be the case that modeling contributes to forecasting, that is not the only—or even the most important—reason to build models. Models help us explain observations, understand system dynamics, illuminate uncertainties, offer options for interventions, set boundaries of parameters and outcomes, discipline our thinking, and identify new questions. These are valuable, tangible results of applying modeling to social behavior research.

The presentations and workshops raised many important themes and questions, many of which are amenable to modeling approaches. Some of the questions to which mathematical and computational modeling is being applied are highlighted below:

• How does the social context influence how the brain functions and vice versa?
• How can we represent personality, situations, beliefs, and knowledge in models?
• How are attitudes learned and changed? How do people learn stereotypes and other categorized knowledge?
• How do people identify their social groups?
• How does information (e.g., medical practices, rumors, knowledge) spread in social networks? How much connectivity and how much exploration of the network are valuable in disseminating information?
• Are people primarily competitors or altruists? In what situations does conflict arise?
• What are the properties of decentralized social systems?
• What are genetic, biochemical, physiological, and environmental components of social behavior?
• What can we learn from studies of model organisms?
• To what extent do our expectations and assumptions about social systems bias research questions and data collection?

The central themes about the role of modeling in social behavior research emerged:

• Integration of knowledge across levels of organization by encouraging communication and collaboration among researchers from different disciplines
• Development of a conceptual and theoretical framework

In particular, the workshops on prejudice and resilience exemplified how knowledge from many levels and many perspectives can be brought to bear on understanding complex social issues. For example, prejudice can be studied as a structural, cultural, social network, individual, or interpersonal phenomenon, all of which are relevant to understanding and addressing disparities in health and access to resources. Resilience deals with the ability of a system (e.g., individual, community, society) to recover from a crisis. The modeling framework for studying resilience comes from ecological modeling and is now being applied to physiological, social, and cultural systems. With the inclusion of many perspectives, our understanding of the underlying theory is also broadening.

Meeting Recommendations

Research

• Study how to model across levels of organization (molecular to cultural)
• Encourage projects that propose to dock models on various levels of organization
• Improve the variety and quality of mathematical and statistical tools for modeling social behavior
• Identify and develop animal models that allow for integrative analyses of the genetic, biochemical, physiological, and environmental components of social behavior
• Develop individual-based model that includes social interactions and that matches macroscopic, observed patterns of behavior
• Develop models that deepen our understanding of causes and interventions of health disparities
• Encourage studies of internet-based communities
• Provide start-up grants to develop cross-disciplinary research programs

Infrastructure

• Develop cross-disciplinary training in modeling and mathematics of social systems
• Facilitate match-making of researchers, data, and methods
• Improve distribution of information on resources and publications
• Facilitate development of common lexicon, such as descriptions of behaviors

Organizing Committee

John Cacioppo
The University of Chicago

Ana Diez-Rouz
University of Michigan

Joshua Epstein
The Brookings Institute

Jessica Flack
The Santa Fe Institute

Steve Frank
University of California, Irvine

Robert Goldstone
Indiana University

Eric Smith
The Santa Fe Institute

Irene Eckstrand
NIGMS Staff Organizer