Summary: How do strangers converge on the same habit or workplace norm? New research shows it’s not simple mimicry or complex calculation. Instead, people follow a two-stage process: they first sample different behaviors, then commit to one once a clear internal threshold is reached.
That threshold is described by the Tolerance Principle, a compact mathematical rule first used to explain how children learn grammar. The same rule also predicts when individuals adopt social conventions and how collective norms can tip or flip.
Key Facts
- Two-stage learning: Individuals explore options probabilistically before making a categorical commitment once enough consistent evidence accumulates.
- The Tolerance Principle: A simple equation that quantifies how much regularity a person must observe before treating a pattern as a rule despite occasional exceptions.
- Shared mechanism: The cognitive process that helps children generalize grammar also appears to drive adults’ adoption of social conventions and organizational norms.
- Predicting tipping points: The model estimates how large a dissenting minority must be to overturn a prevailing norm, offering a mathematical account of social change.
Source: CUNY
A paper in Proceedings of the National Academy of Sciences (PNAS) presents a concise answer to how people form shared social conventions, from everyday habits to workplace culture.
Researchers at the CUNY Graduate Center, the University of Pennsylvania, and Stanford University report that people do not primarily learn norms by blindly copying others or by constantly calculating the statistically optimal choice. Instead, learning follows a predictable two-stage pattern: sampling followed by commitment once evidence crosses a threshold.
The study shows this commitment threshold is captured by the Tolerance Principle, a parameter-free formula originally developed to model how children infer grammatical rules. Applied to social behavior, it predicts when adults will treat a recurring pattern as a rule despite occasional exceptions.
Spencer Caplan, a linguistics professor at the CUNY Graduate Center and co-lead author, explains that people first try different options because they do not want to lock into a pattern that might be a fluke. “Once the observed behavior reaches the threshold of ‘good enough,’ people switch into applying the rule and often stick with it even when they later observe exceptions,” he said.
To test theories of convention formation, the team built computational models representing different learning strategies and compared them with data from coordination experiments. They analyzed prior published studies and ran new experiments in which participants independently attempted to align on shared choices—such as agreeing on a name for an unfamiliar face—while interacting within social networks. Small rewards for matching others’ responses enabled tracking how collective decisions evolved over time.
Across experiments, participants consistently departed from the two dominant assumptions about social learning: simple imitation and optimization. Instead of immediately copying a recent action or always choosing the statistically optimal response, people behaved probabilistically at first, reflecting uncertainty. Their responses then stabilized once accumulated experience passed a mental threshold predicted by the Tolerance Principle.
The threshold-based model not only matched human behavior better than imitation or optimization models but also outperformed Bayesian and other optimization approaches when predicting how people learn under noise and uncertainty. In a preregistered dyadic experiment, the Tolerance Principle best reproduced observed learning rates among participants tasked with inferring nonlinguistic behavioral patterns.
These findings point to a single, simple cognitive mechanism that may underlie learning across domains: from acquiring grammatical rules in childhood to adopting social norms and workplace practices in adulthood. Like a child who generalizes that “-ed” marks past tense while tolerating irregulars such as went, adults seem to apply a rule once it reaches sufficient weight in their experience.
Understanding this process clarifies how social change occurs. Because the model predicts when people commit to a convention, it also estimates the critical mass required for a dissenting minority to reverse it. That provides a quantitative view of tipping points in collective behavior relevant to public health campaigns, organizational change, and the spread of new ideas.
The authors note that future research will explore how these dynamics operate in more complex real-world settings, where identity, status, power, and other contextual factors interact with coordination demands.
Key Questions Answered:
A: People sample before committing. They avoid locking into a pattern that could be a temporary trend. Only when observed behavior exceeds the mathematical Tolerance threshold do they switch from sampling to applying the rule and tend to stick with it despite later exceptions.
A: Yes. Once a norm surpasses the threshold, individuals stop sampling and begin applying the rule. That makes them resistant to conflicting evidence. To flip the culture, a dissenting minority must be large enough to push collective experience back below the threshold.
A: It’s the same principle. Children hear many examples like “walked,” “talked,” and “played” and eventually form the rule that “-ed” marks past tense. Exceptions such as “ran” or “went” are tolerated because the rule has reached sufficient strength in the child’s internal model.
Editorial Notes:
- This article was edited by a Neuroscience News editor.
- The journal paper was reviewed in full by staff.
- Additional context was added by editorial staff.
About this social neuroscience research news
Author: Shawn Rhea
Source: CUNY
Contact: Shawn Rhea – CUNY
Image: The image is credited to Neuroscience News
Original Research: Closed access.
“A Simple Threshold Captures the Social Learning of Conventions” by Douglas Guilbeault, Spencer Caplan, and Charles Yang. PNAS
DOI: 10.1073/pnas.2508061123
Abstract
A Simple Threshold Captures the Social Learning of Conventions
A long-standing question in cognitive and social science is how people learn social conventions from sparse, noisy observations of many different actors without explicit instruction. Dominant theories like imitation and optimization struggle to account for empirical patterns observed in coordination tasks.
Across multiple experiments where participants coordinated behavior while interacting in networks, choices systematically departed from both pure imitation and optimization. Instead, individuals followed a categorical two-stage process: they sampled probabilistically until accumulated information triggered a mental threshold, after which their choices stabilized.
The researchers estimated this threshold using the Tolerance Principle (TP), a parameter-free equation originally developed to model children’s rule learning in language. Threshold-based agents produced social learning dynamics that better matched human data than imitation or optimizing agents and provided a clearer prediction of how a critical mass of dissenters can overturn conventions.
The TP model outperformed a variety of optimization approaches, including Bayesian inference. In a preregistered dyadic experiment that controlled levels of noise in observed signals, TP best reproduced human learning rates. These results support the idea that a simple mathematical threshold underlies individual and collective learning—from grammar acquisition to establishing behavioral norms.