The question
How does behaviour spread through groups?
One employee begins taking a short walk during lunch.
At first, it is simply something they do alone. A colleague sees them leave each day and eventually asks to join. Now two people return from lunch together. Their routine becomes easier to notice, and other coworkers begin discussing whether they should come too.
A few weeks later, several members of the team regularly walk at lunchtime.
The sequence is consistent with behaviour spreading from one person to another, but it does not prove that this is what happened. The weather may have improved. The workplace may have introduced a health initiative. People who already enjoyed walking may have started spending more time together.
A group pattern can develop through influence, shared conditions, selection or several processes at once. Seeing the final pattern does not identify its cause.
Spread begins with a pathway
For behaviour to spread through observation, it must first become available to other people.
That can happen through direct contact, conversation, demonstration, stories, visible consequences or digital displays. A person sees a colleague walking, hears a friend describe a routine or watches someone receive approval for speaking up.
Behaviour being visible in a setting does not establish that a particular person encountered it. Encountering it does not establish attention, and attention does not establish adoption.
A connection between two people creates a possible pathway. It does not prove that relevant guidance travelled through it. Who encounters what depends on relationships, proximity, schedules, access, organisational roles and platform design. Even people seeing the same action may notice different features.
As explained in Attention Selects Information, only part of what is available becomes something a person actively notices. This selection occurs within the broader social environment described in Human Cognition Is Social. Other people continually make actions, expectations and outcomes available—but not equally to everyone.
Learning makes adoption possible, not automatic
When people encounter behaviour, they may learn that an action exists, how it is performed, what goal it serves or what consequences appear to follow. Instructions, reports and visible outcomes can also supply guidance.
These individual processes are examined in Learning from Other People. They provide one foundation for transmission across a group.
But learning does not guarantee adoption. A coworker may understand the lunchtime routine without wanting to join. Someone may admire a colleague who speaks openly but consider the personal risk too high. A person may learn how to use a tool but lack access to it.
Adoption can depend on motivation, capability, opportunity, resources, safety and competing goals. Continued performance also depends on whether the behaviour remains possible and sufficiently supported by its consequences and context.
Spread can stop between any of these stages: encounter, attention, learning, adoption and continued performance.
What people do and what they approve send different signals
Groups communicate at least two kinds of normative information.
A descriptive norm concerns what people appear to do: Most people here answer messages after working hours.
An injunctive norm concerns what people appear to approve, expect or discourage: People here believe committed employees should remain available.
These signals can align, but they do not have to. Many employees might answer messages late while privately disliking the expectation. A community might approve of recycling even though few residents consistently do it. A risky practice might be common while receiving official criticism.
As Norms Make Behaviour Feel Natural explains, repeated visibility can increase familiarity and may contribute to an inference that behaviour is normal. Perceived prevalence may also suggest what is possible or socially safe.
But common behaviour is not necessarily approved, and approved behaviour is not necessarily common. People may misjudge both. A highly visible minority can sometimes be mistaken for a majority, while private disagreement remains hidden.
Learning what a group does or expects does not determine compliance.
Consequences affect whether behaviour remains visible
Continued performance often creates new exposure, although records, stories and digital traces can preserve visibility after the original action ends.
Consequences can affect whether behaviour continues. The lunchtime walk may improve energy, require effort or attract friendly attention. Speaking up in a meeting may lead to recognition, criticism or no visible response.
Some consequences are practical outcomes experienced by the performer. Others are social responses such as approval, criticism, inclusion or exclusion. Observers may also learn from seeing what happens to someone else without receiving the same consequence directly.
The broader architecture of reinforcement and consequences is established in Correlation, Prediction, Causation and Mechanism. Within group spread, consequences can strengthen or weaken later performance and change what observers expect.
But reinforcement does not necessarily explain how behaviour was first acquired, attention is not inherently rewarding and the same response can mean different things to different people.
Belonging changes which examples feel relevant
People do not treat every visible person as an equally relevant model.
In some contexts, a trusted peer may matter more than an unfamiliar expert. Someone in a desired role may demonstrate what a future self could do. A person with comparable circumstances may make an outcome seem more applicable.
Belonging can affect who receives attention, whose consequences matter, which behaviour appears socially safe and whether acquired knowledge is expressed publicly. As explored in Belonging Regulates Behaviour, acceptance and affiliation can give some actions greater motivational significance.
Personal, Social and Role Identity distinguishes the ways personal characteristics, social memberships and roles organise identity. Which membership is active in a situation may affect whose conduct appears relevant.
This remains conditional. Identification does not cause imitation. People can resist a group expectation precisely because their membership matters to them, while outsiders may become important sources when they possess relevant knowledge.
Some positions create more visibility than others
Exposure is not distributed evenly.
Some people interact with more group members or occupy roles that make their actions easier to observe. Authority, audience size and control over resources can increase visibility. Algorithmic or editorial amplification can also make particular people prominent independently of their social status.
Status can affect who receives attention, a broader process examined in Status Changes Attention and Action, the next piece in this sequence.
But visibility, credibility, expertise and influence are different. A person can be highly visible without being trusted. They can hold authority without possessing relevant expertise. A central member may reach many people without changing what any of them do.
A less visible peer may matter more within one trusted relationship. Influence depends on the behaviour, audience, relationship and situation; it is not a fixed property that some people simply possess.
Networks shape who encounters whom
A network is the pattern of connections through which people encounter, observe or communicate with one another.
Its structure affects possible pathways. Someone who interacts with many people may create broad exposure. A tightly connected cluster may repeatedly encounter the same practice. A person connecting otherwise separate groups may carry guidance between them. An isolated member may receive little exposure even when behaviour is common elsewhere.
Different structures matter for different behaviours. Some information can be acquired after one encounter. Some risky or demanding behaviours may be more likely to be adopted after reinforcement from several sources. Clustering can support local adoption while limiting movement beyond that cluster.
Network position creates opportunities and constraints. It does not determine what people notice or do. A recorded connection is also imperfect evidence: two people may be linked without discussing the behaviour, while an unrecorded conversation may provide the actual exposure.
Controlled online studies can show that a network arrangement affects exposure and adoption under specified conditions. They do not establish that the same effect has the same strength in workplaces, families, communities or other platforms.
Groups and cultures also differ in which relationships carry authority, how behaviour becomes visible and what approval or belonging means. A pathway observed in one setting should not be assumed to operate identically elsewhere.
Adoption can change the conditions for later adoption
When another person adopts a behaviour, they can become a new source of exposure.
One possible feedback sequence is:
- behaviour becomes visible;
- some people adopt it;
- adoption creates further exposure;
- visible prevalence changes;
- further adoption becomes possible.
Every transition can fail. One person may act after seeing a trusted friend. Another may wait for several peers. For someone else, official approval or reduced practical risk may make adoption more likely. These are varying adoption conditions, not fixed traits that can always be observed.
The process may become self-reinforcing, a systems pattern developed more fully in Feedback Loops Sustain Patterns, a later piece in the Library. Yet more exposure can also produce fatigue, criticism or rejection. Negative consequences may become visible, and counter-norms may strengthen. Feedback can accelerate, stabilise or stop a pattern.
Similar people may have selected one another
Groups can become behaviourally similar without members changing one another.
People often form or maintain relationships partly because they already share interests, values, occupations, locations, resources or routines. Runners may become friends because they already run. People who prefer late-night work may join the same online community.
This tendency for similar people to connect is called homophily.
If a friendship group contains many runners, friendship may have encouraged additional running. But the pattern may also reflect who became friends in the first place.
Selection and influence can operate together. Prior similarity can contribute to connection, later interaction can deepen similarity, and changing behaviour can alter which relationships continue.
Similarity among connected people therefore does not establish that behaviour spread between them.
Shared conditions can produce the same behaviour
A shared setting can both connect people and expose them to the same conditions.
Coworkers share policies, schedules and facilities. Neighbours encounter the same transport system and local prices. Platform users encounter the same interface and recommendation system.
Suppose several employees begin cycling after their workplace installs secure bicycle storage. The facility changed what was possible for all of them. They may also observe and encourage one another, but interpersonal influence is not the only explanation.
A shared environment is an ongoing context people occupy; a common external cause is an event or change that affects several people. Rain can make many people carry umbrellas. A new policy, economic shock, campaign, news event or platform update can likewise alter several people's behaviour without one member causing another person's response.
These explanations are not mutually exclusive. An environmental change may create opportunity while visible adoption supplies guidance and social support.
A group pattern does not explain itself
When connected people behave similarly, at least four explanations should remain open:
- one person influenced another;
- similar people selected one another;
- members encountered shared conditions;
- the same external event affected several people.
Real patterns may involve all four.
Separating them is difficult because people are not randomly assigned to relationships. Behaviour can change which relationships form or continue. Exposure may remain hidden. Environmental causes may be measured poorly, and influence can run in both directions.
Randomised or credible natural experiments generally support stronger causal conclusions. Detailed longitudinal evidence can improve inference but remains dependent on measurement and assumptions.
A pattern can raise a question about influence. It cannot answer that question by itself.
Popular behaviour is not necessarily useful behaviour
Spread describes prevalence, not value.
A behaviour may spread because it is visible, easy, rewarded, familiar, emotionally striking or repeatedly promoted. None establishes that it is effective, accurate, safe or fair.
Useful practices can spread. So can inefficient routines, inaccurate claims, harmful norms, risky challenges and discriminatory conduct.
Perceived popularity can provide a fallible signal about what other people do. It cannot decide whether their behaviour deserves to be followed. Value must be evaluated separately from transmission success.
Spread can stop, reverse or remain local
Propagation is not inevitable.
A behaviour may fail to spread because few people encounter it, adoption is difficult, opportunity is absent or consequences are unattractive. Competing norms may be stronger, trusted people may reject it or initial adopters may stop before creating further exposure.
Spread can remain inside one friendship group, workplace, role, platform or period. Adoption can decline or a local pattern can reverse direction when costs become visible, incentives change, alternatives appear, institutions intervene or networks reorganise.
Some local practices persist and become part of the cultural processes examined in Culture Stores and Transmits Solutions. Others contribute to broader patterns that no single person planned or controls, as explored in Groups Have Patterns No Member Controls.
This piece stops before those larger processes. Its question is how behaviour moves across connected people in the first place.
Spread is a pathway, not a diagnosis
Several employees now walk together at lunchtime. Observation and encouragement may have mattered. So may belonging, improved weather, a workplace initiative and selection among people who already liked walking.
The visible cluster does not explain itself.
Behaviour spreads when actions become visible, meaningful and available for further adoption. Networks shape the pathways, while norms and consequences shape what exposure means.
But similarity is an outcome to explain—not proof that contagion occurred.