Ernst Strüngmann Forum

José Balsa-Barreiro, Kim A. Bard, Steven Bedrick, Michael Brauer, Thomas Brinkhoff, Nitesh V. Chawla, Tamas Dávid-Barrett, Megan Doerr, Guillaume Dumas, Peter Ejbye-Ernst, Sophia Frangou, Camilla Bank Friis, Jason Gilliland, Kimmo Kaski, Heidi Keller, Fabio Kon, Hye-Chung Kum, Lasse Suonperä Liebst, Marie Rosenkrantz Lindegaard, Gina S. Lovasi, Daniel P. Lupp, Claudia Bauzer Medeiros, Maria Melchior, Mónica Menendez, Virginia Pallante, Tomas Paus, BeateRitz, Sven Sandin, Abeed Sarker, Cason D. Schmit, Lindsey Smith, Kimberly M. Thompson, Henning Tiemeier, Michele C. Weigle

Science is a highly specialized enterprise—one that enables areas of enquiry to be minutely pursued, establishes working paradigms and normative standards, and supports rigor in experimental research. All too often, however, “problems” are encountered that fall outside the scope of any single discipline, and to progress, new perspectives are needed to expand conceptualization, increase understanding, and define trajectories for research to pursue.

The Ernst Strüngmann Forum was established in 2006 to address such topics. Founded on the tenets of scientific independence and the inquisitive nature of the human mind, we provide a platform for experts to scrutinize topics that require input from multiple areas of expertise. Our gatherings, or Forums, take the form of intellectual retreats: disciplinary idiosyncrasies are put aside, existing perspectives are questioned. Importantly, consensus is not necessarily the goal. Instead, participants work to expose gaps in current knowledge and ways to fill these gaps are collectively sought. To ensure access to emerging insights, the results of the entire process are disseminated through the Strüngmann Forum Report series.

This volume reports on the discussions surrounding the topic of “digital ethology” (i.e., the study of human behavior revealed through multifaceted digital footprints). Tomas Paus (Professor of Psychiatry and Neuroscience, University of Montreal) brought this topic to our attention in 2019. Having participated in two earlier forums, Paus was keen to explore how digital ethology might be used as a conceptual framework and tool to quantify the social environment, and what novel insights into the social dynamics of populations might emerge to generate new knowledge about human behavior across various communities. He invited Hye-Chung Kum (Professor of Health Policy and Management, and Computer Science & Engineering, at Texas A&M University) to join him in preparing a proposal. After review and approval by our scientific advisory board, the Program Advisory Committee was formed to transform the proposal into a framework that would support an extended, multidisciplinary discussion. Joining us on the committee were Kimmo Kaski (Dept. of Computer Science, Aalto University) and Maria Melchior (Sorbonne Université, INSERM, Institut Pierre Louis d’Epidémiologie et de Santé Publique). Together, the committee identified participants and formulated the following overarching goals to guide the discussion:

• To expand understanding of how the environment shapes human development across the life span
• To examine ways through which digital data can broaden research into human behavior and support future comparative behavioral studies across species
• To construct a conceptual and methodological framework for integrating various data sources

Further, the committee established four primary areas around which work would focus and invited “background papers” key topics to initiate the discussion. Originally scheduled to take place from September 20–25, 2020, the Forum experienced delays due to travel restrictions associated with COVID. Ultimately, people traveled to Frankfurt from July 24–29, 2022, for the Forum and a lively discussion ensued between experts from geospatial and data science, behavioral and brain science, epidemiology and public health, ethics, and law, as well as urban planning. This volume synthesizes the ideas and perspectives that emerged.

An endeavor of this kind, especially one developed during COVID lockdowns, creates unique group dynamics and puts demands on everyone. I wish to thank each person who participated in the Forum for their time, efforts, and positive attitudes. A special word of thanks goes to the members of the Program Advisory Committee as well as to the authors and reviewers of the background papers. Importantly, the work of the discussion groups’ moderators—Kim A. Bard, Beate Ritz, Jason Gilliland, and Kimmo Kaski—and rapporteurs—Guillaume Dumas, Gina S. Lovasi, Michele C. Weigle, and Claudia Bauzer Medeiros—deserves special recognition: To support lively debate and transform this into a coherent, multiauthor report is never a simple matter. Finally, I extend my sincere appreciation to the scientific chairs, Tomas Paus and Hye-Chung Kum. Their expertise and leadership accompanied the entire project and contributed greatly to its outcome.

The Ernst Strüngmann Forum is able to conduct its work in the service of science and society due to the generous backing of the Ernst Strüngmann Foundation, established by Dr. Andreas and Dr. Thomas Strüngmann in honor of their father. I also wish to acknowledge the support received from our Scientific Advisory Board as well as the Deutsche Forschungsgemeinschaft, which provided supplemental financial support.

In the attempt to extend the boundaries of knowledge, it is never easy to relinquish long-held views or ideas. Yet once such limitations are recognized, the act of formulating strategies to get past this point becomes a most invigorating activity. On behalf of everyone involved, I hope this volume is able to transfer some of this excitement and be used to create a greater understanding of the relationships between human behavior and the environment through their digital footprints.

From conception onward, the individual is developing, maturing, working, playing, and aging in their context. As illustrated in Figure 1.1, multiple layers of environment (context) surround an individual across space and time: from the uteroplacental circulation connecting the fetus and their mother before birth, to the influence of their caregivers, extended family, and peers during childhood, adolescence, and adulthood. This “proximal” context (light gray) is embedded in larger geospatial units, such as specific neighborhoods, cities, or countries (dark gray). All environmental influences unfold in time throughout the individual’s lifespan. Needless to say, the different layers interact, in a bidirectional manner, with each other. Thus, for instance, a pregnant person responds to signals generated by the fetus, and vice versa (Fowden et al. 2022; Kolle et al. 2020; Menon 2019), the pregnant person interacts with their partner, and vice versa (Khaled et al. 2021; Saxbe et al. 2018), and the caregiver interacts with the child, and vice versa (Carollo et al. 2023; Paquette and St. George 2023). At the same time, the individual and those in their proximal context (e.g., caregivers and peers) act as both recipients and co-creators of their area-level environment along all its dimensions, including physical environment (e.g., air quality), built environment (e.g., parks and transportation network), and social environment (e.g., social cohesion). Different aspects of the environment change over time in an interdependent fashion (e.g., air quality, vehicular traffic, lack of green space, demographic characteristics), often reflecting the resources and policies in place at different levels of geospatial granularity (e.g., country, city, neighborhood). Both within and across countries, the lack of environmental justice is reflected in disproportional exposures of marginalized communities to various combinations of adverse environments and, in turn, their combined health effects (Van Horne et al. 2023).

For the complete chapter click on button below.

• PDF of the Chapter

Abstract

The ethological approach is used to study naturally occurring behavior. In the modern world, many such behaviors are connected to, and recorded by, a wide array of digital services (e.g., social networking, information search, closed-circuit television). How can ethological concepts be applied to help us characterize the environment in which humans live? What aspects of the ethological approach can guide us to obtain measures captured directly from digital data generated by our everyday activities? What kinds of models do we need to understand how human behaviors/activities can be inferred from the physical and built environment? This chapter explores the bidirectional nature of these relationships; namely, how individuals create their environment, and how the environment shapes the individual. It discusses how to proceed from observation and data sampling to knowledge extraction and causal inference. The complementary nature of common and specific are addressed as well as the challenge of integrating niches at both physical and social levels. Finally, all these concepts and associated methods are illustrated through a hypothetical study.

• PDF of the Chapter

Abstract

Digital data can be used to observe human behavior as well as aspects of the physical, built, and natural environment that provide context for such behaviors. Data extracted from communities through surveillance have rightfully been the subject of concern, yet such data hold great potential for benefits, including knowledge generation and dissemination to advance human health and equity. Benefits will depend on what is measured and who sets the agenda. Here, ways to organize available and future physical, built, and natural environment measures are discussed, and approaches are proposed to guide the use of such data to generate knowledge while keeping in mind varied value judgments and goals. Metadata are identified as a key tool to deter misrepresentation and misuse of data. To serve this purpose, metadata could be expanded in several ways, including historical context and intent of data collection as well as limitations and permissions to be aware of while planning use and interpreting findings. As data are used, subsequent versions of metadata could record information to inform future use, including a statement of social license updated as the individuals and communities affected by use of the data reflect on harms and benefits. The process of seeking social license for use of geographically referenced data itself has potential to add to our understanding of human agency and to inform ethical inquiry about the structural determinants and individual choices that play out in communities. Opportunities to fill gaps and meet future challenges are identified. Further, attention must be given to incentives across the funding, publishing, and institutional landscape so that envisioned change can be realized and sustained.

• PDF of the Chapter

The study of social environments has typically revolved around interactions in the physical world. Here, a contemporary perspective of social environments weaves together multidisciplinary viewpoints and considers both physical and virtual spaces that offer opportunities for interaction. In the intersection where virtual and physical spaces collide, how does the structure of the social environment in the physical world affect that in the virtual world, and vice versa? How can abundant area-level digital data, produced at multiple locations and points in time, be used to study these social environments? This chapter examines the role that digital data plays in the study of human interactions, with considerations for context, in terms of physical proximity, history, and culture, as well as the advantages and challenges presented in using social media data for this type of study. The long-term goal is to examine how the social environment extends from the physical verse to the metaverse. This provides an unprecedented opportunity to characterize not only social environments using digital data but also to juxtapose them with the influence of physical environments.

• PDF of the Chapter

Abstract

Knowledge integration permeates all scientific endeavors, which increasingly depend on interdisciplinary collaboration as well as on combining data from multiple sources and knowledge domains. Advances in digital ethology progressively rely on knowledge integration, which is enhanced, but also hampered, by the large volumes of heterogeneous data that need to be considered, the multiple aggregation levels to be considered, and the human expertise involved in answering research questions. Though considerable research efforts have focused on leveraging knowledge creation through data integration, many challenges remain. This chapter identifies and investigates some of these challenges, pointing out strategies toward the generation of knowledge while bearing incentives and barriers in mind. To investigate human behavior in the built, social, and/or natural environments, for example, what kinds of considerations exist when integrating individual and population data? Are big data an asset or a hindrance to such integration? Why should (or should not) researchers go through the effort of curating, documenting, and integrating multiscale data?

First and foremost, despite all the technological advances, human judgment remains a key factor in the selection of datasets to be integrated, in monitoring and validating the integration process, as well as in interpreting the results to extract knowledge. Moreover, quality factors, such as reproducibility or robustness, must be considered at all stages: data selection, design and implementation of the integration process, and result analysis. Appropriate documentation of data and processes must be ensured for fairness and reproducibility, and metadata quality is essential for sharing of data and processes. In conclusion, ethical and legal considerations interact in many complex ways, but there exist paths to move forward and overcome the barriers posed.

• PDF of the Chapter

Abstract

Today, large amounts of digital data about human activities are generated and stored in databases. These data are often geospatial (i.e., locations on Earth are directly or indirectly referenced). To analyze the digital footprint of human activities in their environment, geospatial information is essential because spatial (and temporal) proximity to events may indicate meaningful relationships. The processing, analysis, and presentation of such information require a deliberate handling of geospatial data as well as the use of suitable software tools and frameworks. This chapter provides a short review of the geospatial information technology (IT) systems that can be used for digital ethology. It introduces the main concepts of geospatial information, presents several types of IT systems for handling geospatial data, and discusses their suitability for digital ethology. Special attention is given to the handling of very large geospatial datasets, to the use of geospatial analysis and aggregation methods, as well as to the application of comprehensible visualization techniques. Besides the usage of out-of-the-box functions, more complex geospatial analyses may need to use application programming interfaces for specific solutions.

• PDF of the Chapter

Abstract

Processes such as urbanization demonstrate how human activity influences the physical environment and the subsequent implications for Earth’s natural systems. Correspondingly, changes to different environments, and in particular built environments, are linked with human behavior and health. Understanding these relationships requires the definition and measurement of environments. Considering advancements in the collection and processing of high-volume and high-velocity geospatial data, this chapter seeks to outline features of physical and built environments that may be identified from digital data. Attention is given to open data with varying spatial and temporal resolutions. Administrative data, remote sensing imagery, and data from stationary sensors provide contextual information such as the rate of urban expansion and changes in air quality. Mobile and social sensing enable the collection of high-resolution data that contribute to the identification of smaller-scale features. Developments in classification techniques, such as deep learning, provide the opportunity to explore human–environment interactions in real time. Although challenges exist related to data integration and categorization and must be resolved by future research, the combination of data from multiple sources adds value and holds promise for improving our understanding of the patterns that rapidly change landscapes, and the role of environments in shaping human behavior.

• PDF of the Chapter

Abstract

Over the past century, urbanization has witnessed a significant rise, with the global population in urban areas surpassing 55% today and expected to reach nearly 70% by 2050. While cities contribute to productivity and innovation, dense urban living can bring challenges such as increased living costs, social segregation, traffic congestion, and rising levels of air pollution. The COVID-19 pandemic, coupled with technological advancements and social shifts, has reshaped urban landscapes. Since the majority of the world’s population resides in urban areas, addressing societal and environmental challenges necessitates a focus on cities. This chapter explores the intricate relationship between urban form and social behavior, drawing insights from an extensive review of literature across various themes: human cooperation, mobility, social interactions, integration, quality of life, health, and safety perception. These findings provide a comprehensive framework to understand the complexities of social dynamics in urban environments.

• PDF of the Chapter

Abstract

Originating in biology, the ethological approach to studying human behavior has increasingly spread across various disciplines, including the social sciences. In addition to offering biologically proximate and evolutionary explanations, ethology provides a methodological framework for systematically observing and analyzing human behavior in natural face-to-face settings. This chapter discusses the relevance of using the ethological approach for the study of human behavior, particularly by leveraging video recordings of public behavior for ethological observation. This prospect is demonstrated through an outline of recent video-observational research on violent and bystander helping behaviors. Further avenues are discussed to advance video-based human ethology.

• PDF of the Chapter

Abstract

The adoption of social media is currently at an all-time high. More than half of the world has access to social media. The large-scale adoption and growth of social media have demonstrated the benefits and drawbacks of human activities over such platforms. As the digital footprint of human behavior via social media platforms continues to evolve, it is essential to identify strategies and execute actions that can utilize the data generated for the benefit of humankind. Since most of the human footprint on social media is in the form of free text, the field of natural language processing holds substantial promise in converting such data into valuable and actionable knowledge. Geolocation-related metadata available with or inferred from social media posts enable knowledge to be aggregated at various spatiotemporal granularities. Fine-grained area-level insights about human behavior can, for instance, be obtained through social media-based surveillance in close to real time. Geolocation-specific statistics derived from social media data may also be combined with other area-level data from more traditional sources to obtain comprehensive knowledge on chosen topics. Following a brief introduction to social media and natural language processing, the utility of social media data, particularly when combined with geolocation-based information, is discussed. Two examples—COVID-19 and substance use—are used as case studies.

• PDF of the Chapter

Abstract

Modern technologies and societal changes have generated vast amounts of data, personal and individual or aggregated in clusters or geographic regions. Even though this development has stimulated a wealth of research aimed at understanding disease etiologies and promoting lifestyle changes, opportunities remain, and the integration of data is underutilized.

This chapter describes how geographic and aggregate-level data, with information about environmental and social exposures, can be combined with individual-level health data to increase our understanding of disease etiologies. With an emphasis on data primarily available in Nordic countries, it provides a summary of data sources, references for further reading, approaches and methods for analyses, legal aspects, and limitations.

Compared with data at the individual level, analysis of data at the aggregate level has many advantages in terms of access and privacy. Nonetheless, because the availability of individual-level data is the main strength of data from the Nordic countries, the summary starts with a description of these data and ends with aggregate and geographical (area-level) data. Note that in the Nordic countries, all register-based individual-level data can be linked to geographic regions (e.g., hospital, city, county) associated, for example, with place of birth or current residence. The information provided here should be helpful for anyone interested in disease-specific research and public health work to understand better underlying risks and causal paths.

• PDF of the Chapter

Abstract

In today’s digital world, traces of almost all human activity are logged in various databases, which some have termed the social genome data. When appropriate methods are applied to this real-world data, the potential for new insights is endless. The social genome data may transform many fields of science, just as the human genome data has transformed biology. Yet, obtaining, accessing, integrating, cleaning, and using the social genome data to realize its full potential has many computational, statistical, and ethical challenges. The general methodological approach adopted to study human behavior found in the social genome data is data science. The application of data science in an iterative spiral process can result in the transformation of data to information to knowledge to action by iterating between inductive and deductive reasoning. Data science applies methods from both computer science and statistics, and also seeks to synthesize them and develop new methods to address the context and needs of a particular disciplinary field. In this paper, the importance of incorporating human judgment and expert domain knowledge into the data science activities at all steps and the numerous design decisions required to obtain valid results and ultimately useful insights is emphasized. Challenges and open questions in applying data science to the emerging field of digital ethology for scientific inquiry follow. In sum, data science teams must have a wide view to see the context, understand ethical considerations of the data, and be able to communicate both the insights and the limitations inherent in the data.

• PDF of the Chapter