New Study Unveils the Mathematical Threshold Guiding Social Learning and Convention Formation
In a groundbreaking study published April 22, 2026, in the Proceedings of the National Academy of Sciences, researchers from the CUNY Graduate Center, University of Pennsylvania, and Stanford University have challenged long-held assumptions about how people learn social conventions. Their findings reveal a surprisingly elegant cognitive mechanism that governs how individuals move from tentative exploration to firm commitment in adopting shared behaviors, fundamentally reshaping our understanding of social learning across diverse contexts.
For decades, social scientists have debated whether humans primarily learn social conventions by direct imitation—simply copying others—or by carefully calculating the most statistically advantageous option. This novel research demonstrates that neither theory fully captures the complexity of social adaptation. Instead, the study introduces a sophisticated yet straightforward model: individuals initially sample behaviors probabilistically, reflecting an inherent uncertainty, and then decisively commit to a convention once accumulated experience crosses a critical threshold.
This critical threshold corresponds to a mathematical construct known as the Tolerance Principle. Initially formulated to explain children’s acquisition of language, particularly how they discern grammatical rules amidst irregular verbs and exceptions, the Tolerance Principle provides a predictive framework for when a pattern is “good enough” to be internalized as a rule rather than an occasional quirk. Remarkably, it turns out this principle extends beyond linguistics to adult social learning, guiding how people converge on shared behaviors ranging from everyday habits to complex workplace norms.
The research team conducted rigorous computational modeling, pitting various learning strategies against empirical data from coordination experiments. These experiments required participants to independently align behaviors within social networks—such as choosing a name for an unfamiliar face—while receiving incentives for matching others. The controlled design enabled the researchers to trace how behavioral choices evolved through repeated interaction, shedding light on the cognitive calculus that underpins social convention adoption.
Analysis revealed a consistent pattern: participants did not merely imitate the most recent or popular behavior, nor did they always compute the statistically optimal choice based on observed frequencies. Rather, their choices exhibited an exploratory phase marked by probabilistic variability. As experience accrued, however, participants reached a decisive tipping point, after which their adherence to the recognized convention became stable—even in the face of contrary evidence.
At the heart of this shift is the Tolerance Principle’s mathematical formulation, which defines when accumulated observations justify adopting a pattern as a rule-with-exceptions. This elegantly simple equation captures the balancing act between embracing consistency and tolerating deviations. The study’s computational models incorporating the Tolerance Principle provided superior fits to observed human behavior compared to traditional Bayesian or imitation-based models, revealing the principle’s power in forecasting real-world social learning dynamics.
Beyond illuminating individual cognition, the findings offer profound insights into the dynamics of social change. Since the model specifies when people commit to a convention, it also enables predictions about the size and influence required for dissenting minorities to destabilize entrenched norms. This method of quantifying tipping points in collective behavior has sweeping implications for understanding how innovations spread, social norms evolve, or long-standing conventions are overturned.
This conceptual breakthrough reframes normative behavior as a product of a core cognitive mechanism, underlying diverse forms of social learning. Just as children acquire the general “-ed” past tense rule while noting exceptions like “went,” adults appear to internalize social rules with a similar tolerance for irregularities. The researchers posit this shared mechanism provides a unified framework linking linguistic development with social norm formation.
Co-lead author Spencer Caplan, professor of Linguistics at the CUNY Graduate Center, emphasized the human intuition captured by their findings. “People don’t blindly copy or endlessly calculate,” Caplan noted. “Instead, they explore the social landscape, and once a pattern seems reliable enough—good enough—they commit and stick with it, even when some contradictory evidence emerges. This dynamic mirrors basic human cognition across domains.”
The research team’s findings also resonate with broader societal challenges. Understanding how and when conventions stabilize can inform public health interventions, corporate culture shifts, or strategies for promoting sustainable behaviors. By quantifying the threshold at which social change becomes irreversible, policymakers and organizational leaders can better design initiatives that harness these cognitive principles.
Looking ahead, the authors acknowledge complexities yet to be captured. Real-world conventions are entangled not only with coordination but also with intricate social factors such as identity, status, and power dynamics. Future investigations aim to extend this framework to more nuanced social environments, revealing how these additional dimensions modulate the threshold processes identified.
The contribution of this study lies in its unification of social learning theories through a simple, mathematically grounded threshold that mirrors human cognitive strategies. It offers a parsimonious explanation bridging linguistic and social sciences, while providing practical tools to predict and influence social dynamics in an increasingly interconnected world.
In summary, this pioneering work unveiled in PNAS transforms our grasp of how conventions are learned, upheld, and sometimes overturned. By rooting social learning in a threshold mechanism grounded in the Tolerance Principle, the study provides a powerful model explaining the emergence and persistence of shared norms. As social systems continue to evolve rapidly, these insights are poised to have a lasting impact on science, policy, and society.
Subject of Research: People
Article Title: A Simple Threshold Captures the Social Learning of Conventions
News Publication Date: April 23, 2026
Web References:
DOI Link to Article
Keywords: Social learning, social conventions, Tolerance Principle, cognitive thresholds, social norms, behavioral psychology, coordination experiments, social change, computational modeling, language acquisition, social networks
