Research in social psychology consistently shows that men and women differ markedly in their selectivity when choosing partners for casual relationships, with women typically exhibiting more stringent criteria than men. While this robust phenomenon has traditionally been explained through evolutionary or sociocultural frameworks, we propose an alternative mechanism based on network dynamics. Our explanation derives from two fundamental properties of heterosexual matching networks: first, each match must involve one agent from each group (men and women), and second, changes in selectivity within one group directly affect matching opportunities in the other. Using both analytical and numerical methods, we demonstrate how these properties create a feedback loop that amplifies small inherent differences between the two sexes, inevitably driving one group toward high selectivity and the other toward minimal selectivity. This dynamic renders largely irrelevant within-group variation both in relationship goals and attractiveness: even an attractive man who prefers fewer, quality encounters is driven to become non-selective. This mechanism explains observed sex differences in mate selection without requiring evolutionary adaptations or sociocultural forces, though it remains compatible with their influence.

In the competitive market of Online Social Networks (OSNs) used by the popula- tion, explaining why one platform outperforms another, or why users migrate, remains a complex challenge. While existing literature often emphasizes the competitive ad- vantage created by network effects, our research proposes that network protocols - the foundational rules shaping the creation of OSNs and the interactions within them - play a crucial role in why users prefer one platform over another. To substantiate our argument, we employ computer simulations of different network structures, derived from various network protocols. Our findings reveal significant insights; for instance, directed networks can markedly impede the diffusion of information, and the presence of sub-communities is vital for enhancing collective actions. These simulations demonstrate that the nuances of network design can lead to vastly different outcomes, providing a deeper understanding of user behavior and platform dynamics in online social networks.

An influential network-based explanation for the growing political polarization observed in social media platforms focuses on the self-reinforcing opinion dynamics where social influence leads to the formation of two opposing ideological camps. This explanation overly focuses on the tail ends of the ideological spectrum, thereby overlooking the social processes that describe the moderate majority who are not necessarily a herd of politically apathetic, uninformed, bystanders. How, then, can online polarization intensify despite the existence of this moderate majority? In this study, I develop an explanation that focuses on the amplified relational constraints that users with diverse social networks experience inside highly visible, open communication environments stripped of contextual information. Using a communication network dataset of 26M U.S. Twitter users, I first empirically demonstrate the signs of network brokers’ attempts to mitigate such relational constraints through self-censorship, as measured by their more frequent tweet deletions. Then, I build an agent-based model that explores the generative implications of the empirically observed brokerage and self-censorship association for opinion polarization under varying initial conditions and behavioral assumptions. Extending standard opinion dynamics models that incorporate well-understood polarization mechanisms of homophily and social influence, this model introduces agents’ self-expression choices (i.e., self-censorship), determined by ego-alter and alter-alter opinion differences. These self-censorship decisions at the agent level can collectively distort the perceptions about the true opinion distribution of the population, which, in turn, affects each agent’s subsequent opinion change. Simulation results obtained under a variety of network structures and initial opinion distributions show that, contrary to the empirical findings, agents in brokerage positions generally self-censor less frequently. Furthermore, macro polarization tends to only increase marginally when self-censorship is enabled. However, in completely connected networks that resemble the open, context-collapsed online communication environments, these patterns completely reverse – network brokers (i.e., agents whose tie strengths are more evenly distributed) tend to self-censor more frequently and the system exhibits clear increases in polarization over time. In addition, under conditions of self-censorship, the uncensored opinions at equilibrium appear more polarized than the actual opinion distribution across a broad range of initial conditions. These results hold cautionary implications for empirical research of online opinion polarization where the full range of social media users’ true opinions is difficult to observe.

As ethnic diversity increases in Western countries, concerns about negative outgroup attitudes and interethnic polarization are growing. In diverse classrooms, factors such as peer influence, homophily, and attitude-based selection contribute to changing outgroup attitudes. This study explores if and when individual attitude shifts driven by these processes might scale up to group-level polarization in empirical multiethnic school cohorts that initially have a rather positive interethnic climate. Using an agent-based model calibrated with longitudinal network and attitude data, we conduct a series of computational experiments to assess the effects of hypothetical macro-level “shocks,” such as violent incidents or inflammatory political speeches. Our findings indicate that peer influence, when intensified by an external “shock,” can magnify even small outgroup biases, rapidly leading to polarization. This dynamic occurs despite the presence of other behavioral or network mechanisms that might counteract it. We further provide a detailed analysis of how these results emerge from the empirical context and the interacting processes modeled.

Structural inequalities shape the trajectory of disease outbreaks by influencing exposure risks, access to protective measures, and the effectiveness of interventions. Wealth inequality and social segregation determine who can afford to take protective actions, who remains highly connected within transmission networks, and how rapidly an epidemic spreads through different groups. While empirical studies highlight these disparities, there is a need for computational models to systematically examine how structural inequalities drive infection patterns.

In this study, we develop a computational model that integrates epidemic dynamics, network structures, and behavioral decision-making to analyze how inequality affects disease spread. Our model demonstrates that:

(a) Limited self-quarantine ability among low-income groups widens the infection gap between socioeconomic classes, increasing overall disease prevalence.

(b) Social segregation amplifies transmission by restricting interactions across different socioeconomic status (SES) groups, reinforcing pre-existing disparities.

(c) A second wave of infection can emerge when medium- and high-SES groups develop a false sense of safety, leading to premature exposure and renewed outbreaks.

To validate these findings, we analyze empirical network and economic data from 404 metropolitan areas in the United States and examine infection disparities across ethnic and socioeconomic groups in the City of Chicago. Our results confirm that higher segregation is consistently associated with increased overall infection rates and greater inequality in disease burden.

These findings highlight that structural inequalities are not just passive background conditions but active drivers of epidemic dynamics. Reducing segregation and improving access to protective measures can mitigate disparities and slow disease transmission, underscoring the need for policies that address inequality as a core component of epidemic preparedness and response.