Theoretical perspectives of CHES research

We leverage knowledge and theory from a number of disciplines. These include complexity theory, landscape ecology, geographic information science, data science, and domain knowledge from sociology, demography, economics, and so on. At the same time, our group advances the knowledge, theory, and methodology in the above disciplines. With support from our digital 4-D methodology, all these theoretic perspectives render our work to be: 1) multi-disciplinary—with input from social, natural, and engineering sciences; 2) inter-scale—with ranges varying at spatial, temporal, and organizational scales; and 3) cross-type—with use of both quantitative and qualitative data or models. Below are a set of exemplar theories or frameworks we often build on:

Artificial Intelligence informed Agent-based Complex Systems Science

Modern Artificial Intelligence (AI) informed Agent-based Complex Systems (ACS) Science (AI-informed ACS science hereafter) has several origins. First, it is based on traditional complex systems theory (also known as complexity theory or complexity perspective) that has benefited from general systems theory (von Bertalanffy 1968; Warren et al. 1998). Traditional complex systems theory focuses on understanding complex systems (also named complex adaptive systems). Many grand challenges that are besetting humanity—such as climate change, loss of biodiversity, and the COVID-19 pandemic—can be virtually traced to the actions and interactions of multiple autonomous agents that constitute and drive the complex systems. Such complex systems often encompass a (often large) number of entities and subsystems, among which we often observe multiple interactions, nonlinear relationships, feedback, thresholds, time lags, and adaptation. As a result, these features in complex systems can lead to emergent phenomena or outcomes that are not analytically tractable from system components and their attributes alone.

Traditional artificial intelligence (AI) uses machines to simulate human intelligence, suggesting a “natural” solution for most of the challenges mentioned above. Yet AI’s capability to nourish complex systems understanding is limited by the availability of adequate data and data mining approaches—a daunting challenge in earlier times. In parallel with increasing data availability (e.g., big data, non-traditional qualitative data), data science is advancing rapidly and yielding a wide variety of scientific methods, programming tools, and appropriate data infrastructures to handle such data. This advantage substantially escalates AI’s capability to understand or mimic human intelligence, paving an effective way to simulate agent perception, behavior, and interactions with other agents or changes in the environment(s) around them (Gil and Selman 2019).

In this context, the AI-informed ACS Science focuses on the pivotal role of individual actors (i.e., agents), their adaptive behavior, and the artificial intelligence (particularly its subfield of machine learning) approach with an aim to understand agents' behavior and systems level outcomes (An et al in preparation). AI-informed ACS science leverages the advantages of both machine learning and symbolic methods in order to learn and extract symbolic expressions and rules from data. This constitutes a forefront of research in AI (Lamb et al. 2020). As shown by recent advances in natural language processing, mining qualitative data (e.g., ethnography input, social media texts, and other textual sources) promises to reveal the underlying reasons or explanations for a human agent’s behavior, or their stances towards some issues.

AI-informed ACS science will likely become a new systems science in this era of digital industrial revolution. Aided by inductive reasoning, deductive reasoning, abductive reasoning (for social scientists), and their integration, AI-informed ACS science may fulfill its promise of providing generative theories, advancing a new generic systems theory, and giving humanity an effective means to tackle the grand challenges that it faces (An et al 2023).

Readings and References:


An, L., V. Grimm, A. Sullivan, B.L. Turner II., N. Malleson, A. Heppenstall, C. Vincenot, D. Robinson, X. Ye, J. Liu, E. Lindvist, and W. Tang (2023). Agent-based complex systems science in the light of data mining and artificial intelligence. Environmental Modeling & Software, 105713, https://doi.org/10.1016/j.envsoft.2023.105713.

An, L., V. Grimm, A. Sullivan, B.L. Turner II., N. Malleson, A. Heppenstall, C. Vincenot, D. Robinson, X. Ye, J. Liu, E. Lindvist, and W. Tang (2021). Challenges, tasks, and opportunities in modeling agent-based complex systems. Ecological Modeling 457: 109685. doi: 10.1016/j.ecolmodel.2021.109685

Axelrod, R., and M.D. Cohen (1999). Harnessing complexity: organizational implications of a scientific frontier. The Free Press: New York.

Gil, Y., and B. Selman. 2019. A 20-year community roadmap for artificial intelligence research in the US. arXiv.org :arXiv:1908.02624. Lamb, L. C., A. Garcez, M. Gori, M. Prates, P. Avelar, and M. Vardi. 2020. Graph Neural Networks meet Neural-Symbolic Computing: a survey and perspective. arXiv.org :arXiv:2003.00330.

Liu J, T. Dietz, S.R. Carpenter, M. Alberti, C. Folke, E. Moran, A.N. Pell, P. Deadman, T. Kratz, J. Lubchenco, E. Ostrom, Z. Ouyang, W. Provencher, C.L. Redman, S.H. Schneider, and W.W. Taylor (2007). Complexity of coupled human and natural systems. Science: 317:1513-16.

Von Bertalanffy, L. (1968). General System Theory: Foundations, Development, Applications. George Braziller, Inc.: New York.

Warren, K., C. Franklin, and C.L. Streeter (1998). New directions in systems theory: chaos and complexity. Social Work 43: 357–372.

Data Science

Data science aims to derive insight from data, seeking understanding of the mechanism(s) that generate such data. In scientific investigations, data science is powerful in establishing, confirming, and/or rebutting theories that are instrumental to understanding systems of interest. With essential contribution from artificial intelligence, data analytics, and high-performance computing, data science has been leveraged to set up the basic structure of models, calibrate and validate such models including agent-based models, and project system dynamics. There are numerous successful examples in this regard, including convolutional neural networks (CNNs) for classifying or segmenting hyperspectral satellite imagery (Ma et al. 2019) and aerial vehicle imagery (VidalMata et al. 2020), recurrent neural networks (RNNs) for time series forecasting, graph neural networks (GNNs) for capturing the complex spatial patterns underlying graph-structured geospatial data (Zhu and Liu 2018). Also applications have been extended to modeling COVID-19 transmission (Chimmula and Zhang 2020) and information propagation in social networks (Fan et al. 2019).

Data science has been very instrumental to advancing STEM (Science, Technology, Engineering, and Mathematics) and social science disciplines. Particularly, it shows great potential in understanding complex systems (An et al. 2023). According to Crooks and Wise (2013), crowdsourced spatial data can be used to reveal real-world processes and therefore inform agent-based models in relief efforts. In a successful instance, a graph neural network model has been trained to derive the closed-form, symbolic expression of Newton’s law of motion based on the mass, charge, geographic positioning information, and so on of particles in the experiment (Cranmer et al. 2020). In many instances, deep generative models have been employed to synthesize data, realistically represent the real system’s structure, and produce rules that more accurately mimic processes in target complex systems (Klemmer, Koshiyama, and Flennerhag 2019).

Readings and References:


An, L., V. Grimm, A. Sullivan, B.L. Turner II., N. Malleson, A. Heppenstall, C. Vincenot, D. Robinson, X. Ye, J. Liu, E. Lindvist, and W. Tang (2023). Agent-based complex systems science in the light of data mining and artificial intelligence. Environmental Modeling & Software, 105713, https://doi.org/10.1016/j.envsoft.2023.105713.

Chimmula, V. K. R., and L. Zhang. 2020. Time series forecasting of COVID-19 transmission in Canada using LSTM networks. Chaos, Solitons & Fractals 135.

Cranmer, M., A. Sanchez-Gonzalez, P. Battaglia, R. Xu, K. Cranmer, D. Spergel, and S. Ho. 2020. Discovering symbolic models from deep learning with inductive biases. arXiv:2006.11287 [cs.LG]. https://arxiv.org/abs/2006.11287.

Crooks, A. T., and S. Wise. 2013. GIS and agent-based models for humanitarian assistance. Computers, Environment and Urban Systems 41:100–111.

Fan, W., Y. Ma, Q. Li, Y. He, E. Zhao, J. Tang, and D. Yin. 2019. Graph neural networks for social recommendation. The World Wide Web Conference :417–426.

Klemmer, K., A. Koshiyama, and S. Flennerhag. 2019. Augmenting correlation structures in spatial data using deep generative models. In arXiv preprint arXiv:1905.09796.

Ma, A., A. M. Filippi, Z. Wang, and Z. Yin. 2019. Hyperspectral image classification using similarity measurements-based deep recurrent neural networks. Remote Sensing 11 (2):194.

VidalMata, R. G., S. Banerjee, B. RichardWebster, M. Albright, P. Davalos, S. McCloskey, B. Miller, A. Tambo, S. Ghosh, S. Nagesh, Y. Yuan, Y. Hu, J. Wu, W. Yang, X. Zhang, J. Liu, Z. Wang, H.-T. Chen, T.-W. Huang, W.-C. Chin, Y.-C. Li, M. Lababidi, C. Otto, and W. J. Scheirer. 2020. Bridging the gap between computational photography and visual recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence.

Zhu, D., and Y. Liu. 2018. Modelling spatial patterns using graph convolutional networks (short paper). In 10th International Conference on Geographic Information Science (GIScience 2018), Leibniz International Proceedings in Informatics (LIPIcs)., eds. S. Winter, A. Griffin, and M. Sester, 73:1--73:7. Dagstuhl, Germany: Schloss Dagstuhl--Leibniz-Zentrum fuer Informatik.

Landscape Ecology Theory

Landscape ecology is an increasingly recognized discipline, which addresses interactions between landscape pattern and ecological processes with particular attention to the causes and consequences of landscape heterogeneity over varying, usually large, scales. As a discipline cutting across traditional ecology, geography, and social sciences, landscape ecology offers unique insights into causes and measures of landscape patterns, ecosystem processes, disturbance, landscape connectivity, neutral models, restoration of degraded ecosystems, and human interaction with landscape processes and patterns. In particular, the Fragstats software and many landscape metrics offer big help towards better measuring and understanding landscape processes.

When dealing with CHES research questions, we are interested in multiple concepts, measures, and theoretical perspectives from landscape ecology. For instance, how do landscape patterns (e.g., connectivity) affect ecological processes, human decisions, and human-landscape interaction? How does the space-time principle help our understanding of the nature and scale of various disturbances (e.g., from landslide to hurricane) and landscape processes (e.g., from treefall to wildfire to species migration)? These are also topics of a landscape ecology course we offer. An NSF sponsored project illustrates how landscape ecology concepts and methods help undestand CHES patterns, processes, and mutual relatioships.

Readings and References:


Levin, S.A. (1992). The problem of pattern and scale in ecology. Ecology 73:1943-1967.

Lewison, R., L. An, and X. Chen (2017). Reframing the payments for ecosystem services framework in a coupled human and natural systems context: Strengthening the integration between ecological and human dimensions. Ecosystem Health and Sustainability 3(5), 2017, 1335931.

Turner, M.G. (2005). Landscape ecology: What is the state of the science? Annual Review of Ecology, Evolution, and Systematics 36:319-344.

Geographic Information Science

Geographic(al) information science (GIScience) is the research or discipline that studies fundamental data structures and computational techniques to capture, represent, process, and analyze geographic information (Goodchild 1992). Though closely related to geographic information system(s) known as GIS, GIScience is more about the fundamental concepts, principles, theories, and data structures that underlie many GIS software tools. As a subarea of information science that is about geographic or spatial information, , GIScience consists of essential components such as cartography, geovisualization, geodesy, spatial statistics, GIS, remote sensing, and global positioning systems (GPS). Recent advances in cognitive and information sciences also contribute to GIScience.

The CHES research makes use of many GIScience concepts and methods (tools), and the most salient two of them are geographic information system(s) and remote sensing. However our connection to GIScience goes beyond use of GIS, GPS, and remote sensing, but also extends to include space-time analysis (An et al. 2015), 4-dimensional agent-based modeling (An et al. 2005, 2020), and extending traditional non-spatial methods to spatial data or space-time data analysis (An et al. 2008, 2016). In particular, we borrow and extend metrics from other disciplines to advance our understanding and representation of landscapes or landscape changes. Below is a list of articles that offer basic understanding of GIScience.

Readings and References:


An, L., J. Mak, S. Yang, R. Lewison, D.A. Stow, H.L. Chen, W. Xu, L. Shi, and Y.H. Tsai (2020). Cascading impacts of payments for ecosystem services in complex human-environment systems. The Journal of Artificial Societies and Social Simulation (JASSS) 23/1/5, Special issue.

An, L., M. Tsou, B. Spitzberg, J.M. Gawron, and D.K. Gupta (2016). Latent trajectory models for space-time analysis: An application in deciphering spatial panel data. Geographical Analysis 48 (3): 314–336 (http://dx.doi.org/10.1111/gean.12097).

An, L., M. Tsou, S. Crook, B. Spitzberg, J.M. Gawron, and D.K. Gupta (2015). Space-time analysis: Concepts, methods, and future directions. Annals of Association of American Geographers 105(5): 891-914.

An, L., and D. G. Brown (2008). Survival analysis in land-change science: integrating with GIScience to address temporal complexities. Annals of Association of American Geographers 98(2): 323-344.

An, L., M. Linderman, J. Qi, A. Shortridge, and J. Liu (2005). Exploring complexity in a human-environment system: an agent-based spatial model for multidisciplinary and multi-scale integration. Annals of Association of American Geographers 95 (1): 54-79.

Duckham, M., M.F. Goodchild, and M. Worboys (2004). Foundations of Geographic Information Science. Taylor & Francis

Goodchild, M. (1992). Geographical information science. International Journal of Geographical Information Systems 6 (1): 31–45.

Coupled Human And Natural Systems (CHANS) Framework

Human-nature systems used to be studied largely either in separation or with unidirectional connections: when human systems were studied, they were considered to be constrained by, or with input from/output to, natural systems—put another way, natural systems were only considered as context or background. On the other hand, human systems were often viewed as exogenous influences when studying natural systems. This disciplinary chasm, in parallel with unidirectional connections between natural and human systems, has been shown unable to explain many complexity features (e.g., feedback, nonlinearity and thresholds, heterogeneity, time lags) in human-nature systems.

The coupled human and natural systems (CHANS) are integrated systems in which people interact with natural components. The CHANS concept has evolved in parallel with many closely related concepts, including coupled natural and human (CNH) systems (Liu et al. 2007, An et al. 2014), human-environment systems (Turner et al. 2003), social-ecological systems (SES; Ostrom 2009), and social-environmental systems (Eakin and Luers 2006). The CHANS framework addresses complex interactions and feedback between human and natural systems, which necessitates inclusion of biophysical/ecological variables and human variables, participation of biophysical/ecological and social scientists, and use of tools and techniques from multiple biophysical and social sciences (An et al. 2014, 2020)

Readings and References:


An, L., J. Mak, S. Yang, R. Lewison, D.A. Stow, H.L. Chen, W. Xu, L. Shi, and Y.H. Tsai (2020). Cascading impacts of payments for ecosystem services in complex human-environment systems. The Journal of Artificial Societies and Social Simulation (JASSS) 23/1/5, Special issue.

An, L., A. Zvoleff, J. Liu, and W. Axinn (2014). Agent based modeling in coupled human and natural systems (CHANS): Lessons from a comparative analysis. Annals of Association of American Geographers 104(4):723-745.

An, L. (2012). Modeling human decisions in coupled human and natural systems: review of agent-based models. Ecological Modelling 229(24): 25-36.

Eakin, H., and A. L. Luers (2006). Assessing the vulnerability of social-environmental systems. Annual Review of Environment and Resources 31 (1):365–394.

Liu J, T. Dietz, S.R. Carpenter, M. Alberti, C. Folke, E. Moran, A.N. Pell, P. Deadman, T. Kratz, J. Lubchenco, E. Ostrom, Z. Ouyang, W. Provencher, C.L. Redman, S.H. Schneider, and W.W. Taylor (2007). Complexity of coupled human and natural systems. Science: 317:1513-16.

Liu, J., T. Dietz, S.R. Carpenter, C. Folke, M. Alberti, C.L. Redman, S.H. Schneider, E. Ostrom, A.N. Pell, J. Lubchenco, W.W. Taylor, Z. Ouyang, P. Deadman, T. Kratz, and W. Provencher (2007). Coupled human and natural systems. AMBIO: A Journal of the Human Environment 36(8):639-649.

Ostrom, E. (2009). A general framework for analyzing sustainability of social-ecological systems. Science 325, 419–422 (2009).

Turner, B.L., P.A. Matson, J.J. McCarthy, R.W. Corell, L. Christensen, N. Eckley, G.K. Hovelsrud-Broda, J.X. Kasperson, R.E. Kasperson, A. Luers, M.L. Martello, S. Mathiesen, R. Naylor, C. Polsky, A. Pulsipher, A. Schiller, H. Selin, and N. Tyler (2003). Illustrating the coupled human–environment system for vulnerability analysis: Three case studies. Proceedings of the National Academy of Sciences 100 (14):8080–8085.

Telecoupling Framework

Telecouplings indicate socioeconomic and environmental interactions between two or more areas over (often relatively large) distances (Liu et al. 2013, 2015), and such interactions may take the form of labor migration, tourism, consumption of goods manufactured and transported from afar, and so on. This framework is intellectually connected to the theory of weak ties and social network analysis (Granovetter 1973), which suggest weak ties may translate to strong results at the macro-level under certain conditions such as through connecting groups (Friedkin 1982; Yakubovich 2005).

This integrated framework aims to account for and internalize many socioeconomic and environmental externalities (spillover effects) across space and over time in an increasingly connected world. The framework consists of five major cross-related components: 1) multiple coupled human and natural systems (CHANS), 2) flows of material, information, and energy among systems, 3) agents that facilitate the flows, 4) causes that drive the flows, and 5) consequences that result from the flows. Many CHES systems exactly bear these components, and we are integrating it with other related theories, frameworks, and methods to address many pressing topics such as hazards mitigation & recovery, human decision making and response, landscape planning, and ecosystem restoration.

Readings and References:


Friedkin, N. (1982). Information flow through strong and weak ties in intraorganizational social networks. Social Networks 3: 273–285.

Granovetter, M.S. (1973). The strength of weak ties. American Journal of Sociology 78: 1360–1380.

Liu, J., V. Hull, M. Batistella, R. DeFries, T. Dietz, F. Fu, T. W.Hertel, R. C. Izaurralde, E. F. Lambin, S. Li., L. Martinelli, W. McConnell, E. Moran, R. Naylor, Z. Ouyang, K. Polenske, A. Reenberg, G. de Miranda Rocha, C. Simmons, P. Verburg, P. Vitousek, F. Zhang, and C. Zhu. (2013). Framing sustainability in a telecoupled world. Ecology and Society 18(2):26.

Liu, J., V. Hull, J. Luo, W. Yang, W. Liu, A. Vi?a, C. Vogt, Z. Xu, H. Yang, J. Zhang, L. An, X. Chen, S. Li, W. McConnell, Z. Ouyang, W. Xu, and H. Zhang (2015). Multiple telecouplings and their complex interrelationships. Ecology and Society 20(3):44.

Yakubovich, V. (2005). Weak ties, information, and influence: How workers find jobs in a local Russian labor market. American Sociological Review 70: 408–421.

Multiphasic Response Theory

An important theoretical perspective in our CHES research is to leverage and advance the theory of the multiphasic response (David 1963). This theory hypothesizes that in the early stages of the demographic transition, households had lower child mortality while fertility was still high, leading to increasing population pressures. Therefore these households had to respond in various ways, including deferring marriage, reducing marital fertility, or out-migration. Taking one response would decrease the chances of taking other alternative responses. This was the situation of Japan and Europe in the 19th century. This theory was later expanded to include economic responses, such as expanding the land area, intensifying agricultural production, or finding off-farm work (Bilsborrow 1987). Therefore facing population pressures, people in developing countries will not only respond through demographic transition, but also through economic responses and technological adaptations.

In one of our CHES projects, we are innovatively testing the theory. At the micro-level (e.g., households), more than two multiphasic responses—e.g., out-migration, off-farm work, and intensification of agriculture—are being placed in multilevel simultaneous equations models (Yang et al. in revision). Our findings shed light into what demographic, economic, and/or technological responses will be implemented under what conditions.

Readings and References:


Bilsborrow, R. 1987. Population pressures and agricultural development in developing countries: A conceptual framework and recent evidence. World Development 15(2): 183-203.

Davis, K. 1963. The theory of change and response in modern demographic history. Population Index 29(4): 345-366.

Yang, S. R. Bilsborrow, L. An et al. (in revision). What influences decisions of local people to out-migrate under Payments for Ecosystem Services (PES)? Evidence from a Nature Reserve in China.

Spatial Demography

Demography is the discipline of population studies—in a broader sense, populations of any species could be targets to demographic studies; though human populations are often the "default" subjects. Demographers study the size, structure, distribution, changes, and many other characteristics of populations (e.g., birth, migration, ageing, and death), along with the economic, social, cultural, and biological contexts or processes that exert influences on population processes.

As time goes on, demography has been developing an increasingly explicit awareness of spatial variation and its importance towards demographic studies. In addition to some universal principles, spatial variation may also play an important role in explaining demographic characteristics and/or transitions. Spatial analysis is not only essential for demographic theory development, but also for empirical studies. This is the essence of spatial demography, which focuses on the spatial analysis of demographic processes. In our CHES research, we focus on envisioning or modeling low level (e.g., individual level), spatially-variant population processes (such as birth, marriage, and death) or features (e.g., health outcomes) and how contextual factors may affect them in a spatially explicit manner (Liu et al. accepted). Below we list books, book chapters, and papers that help generic understanding of spatial demography.

Readings and References:


An, L., W. Yang, and J. Liu (2016). Demographic decisions and cascading consequences. Book chapter in Liu et al.: Pandas and People: Coupling Human and Natural Systems for Sustainability. Oxford, UK: Oxford University Press.

An, L., M. Linderman, Guangming He, Z. Ouyang, and J. Liu (2011). Long-term ecological effects of demographic and socioeconomic factors in Wolong Nature Reserve (China). In R.P. Cincotta & L.J. Gorenflo (Eds.), Human Population: Its Influences on Biological Diversity. Berlin, Germany: Springer-Verlag.

An, L., G. He, Z. Liang, and J. Liu (2006). Impacts of demographic and socioeconomic factors on spatio-temporal dynamics of panda habitats. Biodiversity and Conservation 15:2343-2363.

Crook, S.E.S., L. An, D.A. Stow, and J.R. Weeks (2016). Latent trajectory modeling of spatiotemporal relationships between land cover and land use, socioeconomics, and obesity in Ghana. Spatial Demography 4(3):221-244.

Howell, F.M., J.R. Porter, and S.A. Matthews (2016). Recapturing Space: New Middle-Range Theory in Spatial Demography (Spatial Demography Book Series Volume 1). Springer. ISBN: 978-3-319-22809-9 (Print); 978-3-319-22810-5 (Online).

Liu, Y., Dai, J., S. Yang, R. Bilsborrow, M. Wang, and L. An (accepted). Measuring neighborhood impacts on out-migration from Fanjingshan National Nature Reserve, China. Spatial Demography.

Weeks, J.R., D. Stow, and L. An (2018). Demographics, health drivers & impacts on land cover and land use change in Ghana. Chapter for Stephen J. Walsh (ed.), Remote Sensing Applications for Societal Benefits (Comprehensive Remote Sensing Vol. 9), Elsevier.

Weeks, J.R. (2015). Population: Introduction to Concepts and Issues (Twelfth Edition). Boston, MA: Cengage Learning.

Zvoleff, A., and L. An (2014). The effect of reciprocal connections between demographic decision making and land use on decadal dynamics of population and land use change. Ecology and Society 19(2):31.

Zvoleff, A., L. An, J. Stoler, and J.R. Weeks (2013). What if neighbors' neighborhoods differ? The influence of neighborhood definition on health outcomes in Accra. In J.R. Weeks & A.G. Hill (Eds.), Spatial Inequalities: Health, Poverty and Place in Accra, Ghana. Springer.

Our theoretical support is not limited to these perspectives. As an evolving process, our research will leverage a broader knowledgebase, including potential contributions from computer science and engineering, spatial semantics, computational linguistics, human (cultural) geography, political science, and the like. We look forward to new perspectives or theories from more researchers, practitioners, and/or projects.