Welcome

This is my PhD in fun mode. It combine data collection, analysis, model building, narrative, and visualizations under the same roof, with a pinch of 'choose your own adventure' for people who get bored. See my main website for more details

Overview

Humans' unique ability to solve collective action problems arises from the interplay between complex institutions and our cultural psychology [@boyd_culture_1988; @ostrom_governing_1990; @north_institutions_1990; @henrich_weirdest_2020]. Consider, for instance, our ancestors on the Great Plains of North America. For at least the last 10,000 years, hundreds of people have gathered to hunt bison---animals weighing nearly a metric ton and known for their agility and strength, as well as their poor eyesight [@boyd_largescale_2022]. These communal hunts are of particular interest to anthropologists because they involved high-risk and high-reward strategies in which diverse sets of individuals had to cooperate and play specific roles to succeed.

As the scale and complexity of collective challenges expanded---from communal hunts to stateless warfare to managing contagion in industrial societies---humans innovated more elaborate sets of rules, norms, and conventions, collectively referred to as institutions [@north_institutions_1990]. Institutions structure every aspect of human communal life: maintaining peace, providing care, enforcing justice, and managing intergroup relations. Critically, these institutional frameworks co-evolved alongside human cultural psychology, producing minds finely tuned to group cooperation and fairness---traits observable even in young children. Remarkably, even 3- or 4-year-old children demonstrate a deep concern for fairness in group-based problem-solving tasks [@engelmann_childrens_2019].

Complex systems studies have advanced our understanding of social phenomena like contagion, innovation, and cooperation [@newman_networks_2010]. By employing mathematical models that reveal mechanisms behind emergent behaviors such as tipping points, scaling laws, and community formation [@newman_structure_2003; @pastor-satorras_epidemic_2001; @barabasi_emergence_1999]. However, despite their insights, traditional network models often inherit a physics-based view of individuals as mindless entities interacting in different spaces. While useful, this approach falls short of capturing the uniquely human, co-evolutionary dynamics between individuals and institutions.

Institutions are inherently collective phenomena---group-level features that emerge from our unique capacity for sophisticated, non-trivial social interactions. Recent work on higher-order networks (HONs) partially addresses this by modeling group interactions as involving three or more agents, distinct from the independent influence of pairwise ties [@benson_higher-order_2016; @battiston_physics_2021]. Still, this approach often centers on nonlinear effects within groups, overlooking the broader institutional contexts in which such interactions unfold. That is, there are group states beyond higher-order interactions, which are themselves the product of their own dynamics. More generally, before effectively modeling institutions, we must refine our understanding of group interactions themselves. Only then can we better represent the interaction between individual psychology and different aspects of group interactions.

Networked, group interactions

The study of higher-order interactions has exploded in recent years to explore different types of group interactions and their impact on system dynamics [@majhi_dynamics_2022; @ferraz_de_arruda_contagion_2024]. While many models focus on the nonlinearity of within-group dynamics---e.g., quorum effects or threshold contagion---they often overlook other aspects essential to collective action problems. For example, in the bison hunts of the Great Plains, coordination among hunting parties required not only internal role differentiation but also intergroup cooperation and alignment. Thus, group-level behavior is shaped by interactions between groups; group dynamics do not operate in isolation. Theories of collective action make clear that what happens within one group can strongly depend on what happens in others. Capturing such dependencies requires not just higher-order structure but mechanisms that treat groups themselves---not just their members---as coupled, interacting entities.

Institutions co-evolve with individuals

HONs models often overlook the normative contexts in which group dynamics unfold. For example, while they capture how co-authorship patterns of varying orders differ from independent pairwise collaborations, they rarely consider where and how such group interactions originate [@benson_higher-order_2016; @patania_shape_2017; @benson_simplicial_2018; @alvarez-rodriguez_evolutionary_2020; @st-onge_influential_2022]. By bringing institutions into focus---as group-level features that persist beyond transient multi-way interactions---we open the door to new questions: for instance, how have co-authorship norms evolved in response to external pressures to publish more [@smaldino_natural_2016; @tiokhin_shifting_2024]? By distinguishing between group states (e.g., publishing norms adopted by research groups) and individual behaviors, we can better examine how individuals respond to emerging institutional expectations---recognizing variation in those responses. To be clear, we do not suggest that modeling co-authorships as group interactions is incorrect. Rather, we argue that these interactions are embedded in broader institutional dynamics, which deserve greater attention through the lens of complex systems modeling.

Theory and methodological frameworks

đź’ˇ Defining
🗺️ Modeling
đź“Š Measuring

The theoretical framing of this thesis sits at the intersection of three disciplines, namely cultural evolution theory, the philosophy of social ontology, and complex systems. In doing so, we seek to integrate insights from different perspectives on groups. Cultural evolution provides an answer to why and where institutions come from, and why it makes sense that they persist over generations, while social ontology explains why the phenomenology of groups goes beyond what is typically assumed by traditional physics-based models. Taken together, these dimensions provide a nuanced view of the multi-scale co-evolution of individuals and groups, without reducing one level to the other.

Cultural evolution theory: Cultural evolution is essential in explaining how shifting our view from simple networks to a group-based perspective can allow institutions to be perceived as having a degree of independence from their constituents. Cultural evolution theory studies culture as a distinct system of inheritance and selection, capable of producing group-level behaviors and adaptations [@boyd_culture_1988; @cavalli-sforza_cultural_1981]. It emphasizes the human capacity for cumulative cultural evolution, also known as the ratchet effect, through which knowledge builds over generations, enhancing group fitness in local environments [@tomasello_why_2009; @tomasello_cultural_1999; @tennie_ratcheting_2009]. In humans, cultural traits have co-evolved with biology over at least two million years in a process known as gene--culture coevolution. This has embedded culture into both our environment (i.e., the cultural niche) and our biological makeup [@laland_darwins_2018; @tomasello_natural_2014; @laland_how_2010]. As such, cultural evolutionists are interested in modeling the mechanisms that explain group-level cultural diversity, typically in traditional societies [@mcelreath_shared_2001].

A key aspect of cultural psychology is how we coordinate with group members, teach, and learn from each other [@reader_social_2002; @henrich_secret_2016]. We use social learning strategies to decide when and from whom to learn [@laland_social_2004]. When individual learning is costly or outcomes are uncertain, it may be more effective to copy the majority or those who have succeeded in a task. Conversely, in rapidly changing environments, it can be wiser to learn by trial and error, since others' knowledge may be outdated.

Beyond social learning strategies, we are characterized by over-imitation, i.e. the tendency to copy both relevant and irrelevant actions when learning from others [@henrich_evolution_1998; @henrich_evolution_2001]. In various context, humans---especially children---have been shown to replicate an adult demonstrator's exact actions, even when a more efficient method exists or when imitation leads to failure [@nielsen_overimitation_2010]. By acquiring an individual psychology that is overly reliant on learning from each other, we have become a species unique in our ability to trump our own awareness of failure [@schmidt_young_2016; @tomasello_differences_2023]. This tendency to adopt and enforce rules based on others' behaviors, even though it might be individually costly, is the hallmark of scaffolding institutions at the group level.

Social ontology of groups: This thesis also draws from social ontology to explore the role of collective intentionality in shaping group behaviors [@wittgenstein_philosophical_1953; @searle_construction_1995; @gilbert_walking_1990; @jankovic_routledge_2017]. In phenomenology, intentionality refers to the "about-ness" of human experience---our capacity to direct thoughts toward objects or concepts---distinguishing purposeful action from mere stimulus-response behavior [@zahavi_phenomenology_2018; @gallagher_phenomenological_2013; @merleau-ponty_phenomenology_1945]. Collective intentionality, then, refers to the shared "we" that underlies the emotional and cognitive responses individuals make on behalf of their group. Children, once again, excel at passionately enforcing made-up rules, even those created moments earlier during an experiment [@tomasello_differences_2023].

In experimental games, they begin enforcing social norms as early as age 3 or 4---not because the rules matter for task success, but because of their shared sense of "you have to play it this way" [@engelmann_childrens_2019]. As with over-imitation, they might protest against infringed socially constructed rules, even though the infringement does not affect them at all [@schmidt_young_2012]. It is the foundation of culture, where later on people will likewise express moral outrage or indignation when, say, other ethnic groups might exhibit different norms than those that are local customs.

We draw one key insight from the study of intentionality relevant to our work; there is something essential about how individuals experience the 'we-ness' of groups that defines group membership itself. Prominent philosophers of social ontology---and many cultural evolutionists---have argued that humans derive meaning through their participation in groups, not just through individual reasoning. In the thesis conclusion, we refer to this foundational challenge---namely, that group interactions may depend on the mutual recognition of group membership---as the hard problem of intentionality.

Methodological Framing

Our methods draw primarily from the physics of HONs, cultural evolution theory (i.e., population biology applied to cultural systems), and evolutionary game theory--particularly models of cultural diffusion in N-player games [@boyd_origin_2005; @mcelreath_mathematical_2007; @smaldino_modeling_2023; @bowles_cooperative_2011]. More specifically, we use mean-field theory, which plays a central role and recurs throughout the thesis [@kadanoff_more_2009; @guerra_annealed_2010]. Game theory and concepts of bounded rationality are incorporated when integrating group-level models, including institutional dynamics, with individual preferences [@von_neumann_simple_1944; @smith_evolution_1982]. We adopt evolutionary group dynamics as a framework for studying institutions, applying the concept of fitness from evolutionary game theory at the group level (e.g., replicator dynamics). Finally, we use a group-based formalism that builds on recent work with approximate master equations--a description of complex networks that captures detailed within-group dynamics while approximating between-group interactions via mean-field methods [@hebert-dufresne_propagation_2010; @osullivan_mathematical_2015; @st-onge_master_2021].

Our ability to cooperate in large-scale collective action problems is only as complex as the problems we create ourselves. HONs models improve our understanding of group-based phenomena in all sorts of contagion---technological innovations, gossip, cultural norms, and viruses. But each of those contagion processes is equally shaped by our collective, institutional abilities to act together to influence the direction they will take. Thus, this thesis explores the importance of the co-evolution between institutions and individuals. By integrating insights from different disciplines, we arrive at a novel group-based formalism in complex systems that is informed by both cultural evolution and the social ontology of groups. Through modeling group states, we aim to better capture the unique intricacies of human cultural psychology and the role of institutions as evolving group-level features.

Road-map of thesis