Data-informed modeling of the formation, persistence, and evolution of social norms and conventions

TL;DR

The chapter addresses how social norms and conventions form, persist, and evolve, and how quantitative data can inform mathematical models. It surveys population and agent-based modeling approaches, including linear threshold, evolutionary dynamics, and game-theoretic frameworks, with data integration shaping model design and validation. Key contributions include illustrating data-informed threshold estimation, experimental validation of tipping-point predictions, and outlining challenges and opportunities for implementing norm-diffusion theory to real-world issues such as sustainability and emergency response. Collectively, this work highlights the potential of data-driven, theory-guided modeling to inform policy and interventions while acknowledging the need for high-quality data and interdisciplinary collaboration.

Abstract

Social norms and conventions are commonly accepted and adopted behaviors and practices within a social group that guide interactions -- e.g., how to spell a word or how to greet people -- and are central to a group's culture and identity. Understanding the key mechanisms that govern the formation, persistence, and evolution of social norms and conventions in social communities is a problem of paramount importance for a broad range of real-world applications, spanning from preparedness for future emergencies to promotion of sustainable practices. In the past decades, mathematical modeling has emerged as a powerful tool to reproduce and study the complex dynamics of norm and convention change, gaining insights into their mechanisms, and ultimately deriving tools to predict their evolution. The first goal of this chapter is to introduce some of the main mathematical approaches for modeling social norms and conventions, including population models and agent-based models relying on the theories of dynamical systems, evolutionary dynamics, and game theory. The second goal of the chapter is to illustrate how quantitative observations and empirical data can be incorporated into these mathematical models in a systematic manner, establishing a data-based approach to mathematical modeling of formation, persistence, and evolution of social norms and conventions. Finally, current challenges and future opportunities in this growing field of research are discussed.

Figures (8)

• Figure 1: Examples of formation, persistence, and evolution of social norms and conventions. In panel (a), relative use of "sólo" (blue) and "solo" (orange) in written Spanish from Google Ngram data Michel176. In panel (b), map of Europe illustrating the hand used for wedding ring in different countries: blue for vast majority of left hand, orange for vast majority of right hand, gray for regional/cultural differences within the country.
• Figure 2: Two exemplary trajectories of the adoption curve obtained with the discrete-time Bass model in Eq. (\ref{['eq:bass']}) with (a) $p=0.001$ and $q=0.01$; and (b) $p=0.001$ and $q=0.1$. Both plots reproduce an S-shaped curve, with different initial slope and velocity to reach the inflection point.
• Figure 3: Example of a network structure, with node set $\mathcal{V}=\{1,2,3,4,5,6\}$ and edge set $\mathcal{E}=\{(1,2),(1,4),(1,5),(2,5),(3,6),(4,5),(5,6)\}$. Note that node $5$ (in blue) has three neighbors ($\vert \mathcal{N}_5\vert=3$), namely $\mathcal{N}_5=\{2,4,6\}$. Hence, its state $x_5(t)$ evolves as a function of its own current state, and the states of its three neighbors.
• Figure 4: Example of an iteration of the evolutionary dynamics in Axelrod1997. Individual $i$ is chosen and has three neighbors: $1$ and $2$ have a single entry of the state in common with $i$ (first and last, respectively), while $3$ has two entries in common (first and second). Assuming that neighbor $2$ is chosen at random (in green), then, $i$ imitates $2$ by copying one of the three entries in which they differ (selected at random), with probability equal to the fraction of entries that they have in common (i.e., $\frac{1}{4}$), ultimately yielding the update rule written on the right of the figure.
• Figure 5: Example of a network coordination game. Individual $i$ has five neighbors: three of them are currently adopting the status quo (in orange), and two the innovation (blue). Hence, $u_i(0,{\bf x})=3$ and $u_i(1,{\bf x})=2+2\alpha$. Clearly, $i$ would get a larger payoff for adopting the innovation if and only if $\alpha>\frac{1}{2}$.