RESEARCH DOSSIER
Social Learning and Norm Acquisition
Abstract. Social learning and norm acquisition constitute the processes through which individuals acquire actions, conventions, expectations, and practical rules from other agents and social environments. This fundamental critical narrative synthesis traces experimental modeling from Bandura's early studies to contemporary research on imitation, emulation, normative protest, social-learning strategies, cultural transmission, species comparisons, network diffusion, emerging transmission-chain evidence from 2025, and digital social cues. It evaluates the separation between acquiring and expressing a response, the effects of model selection and observed consequences, the distinctive inference from conventional demonstrations to ought-like expectations, and evidence that transmission depends on development, social relationships, cultural setting, and network structure. Laboratory and field experiments, cultural-evolution models, and methodological reviews are compared for what they establish and what remains uncertain. The focus is the acquisition pathway; the representation and immediate behavioral effects of current norms form a connected but independent research domain. The monograph is an integrative assessment of published research, without a new experiment, registered systematic-review protocol, or original mechanistic discovery.
Keywords: social learning; norm acquisition; observational learning; modeling; vicarious reinforcement; imitation; emulation; socialization; cultural transmission; conventional norms; norm enforcement; social learning strategies; majority bias; prestige bias; cumulative cultural evolution; human–artificial learning
The Scientific Object of Social Learning and Norm Acquisition
Social learning is experience-dependent change in an individual's behavior or knowledge that is influenced by information obtained from other agents. Norm acquisition is the more specific process through which a learner acquires expectations about socially recognized ways of acting, including conventions, prescriptions, prohibitions, roles, and circumstances in which a rule applies. Research on this subject studies the transformation from witnessed action to retained representation and from representation to later performance. The relevant observations include matching behavior, intentional selection among demonstrated alternatives, expressed rule judgments, correction of violations, and transmission to further learners. A learned action can become a norm-governed practice when participants treat its form as socially appropriate and expect others to follow it Kendal et al. (2018) Rakoczy et al. (2008).
The primary scientific question is how witnessing others changes what a person knows, expects, does, and regards as obligatory. Controlled modeling studies isolate exposure to a demonstrator; cultural transmission experiments examine how learned practices persist or transform through repeated learners; developmental research asks when children infer prescriptive significance; and comparative studies investigate transmission beyond humans. Each paradigm targets a different dependent variable. A laboratory participant's immediate reproduction of an act provides evidence of performance under those instructions. A later response in a new context is evidence of generalization. Spontaneous correction of another participant's action provides additional evidence of rule-like interpretation. The distinctions matter because the same visible copying can follow instrumental inference, social affiliation, habitual response, or endorsement of convention Over & Carpenter (2012) Mesoudi & Whiten (2008).
This monograph integrates historical experimental studies, behavioral and developmental research, cultural-evolution models, comparative evidence, and research on social networks. It examines what is learned, who is copied, why transmission occurs, how norms are inferred, and how claims about learning can be validated across settings. The object's center is acquisition through observation, interaction, and successive transmission. The current distribution of descriptive and injunctive expectations, and the immediate effects of particular norms on a decision, are treated as necessary conceptual neighbors rather than substituted for acquisition itself. This distinction permits social learning to be studied as a causal pathway connecting individual psychological processes with the historical reproduction of social practices Boyd & Richerson (1985) Bicchieri & Xiao (2009).
Acquisition, Performance, and Normative Understanding
The acquisition of a response and its expression under a test condition are separate experimental outcomes. A learner may observe and encode an action while withholding it because the situation offers little incentive, the model was punished, the setting is socially risky, or an alternative behavior seems preferable. Conversely, immediate behavioral matching can arise from short-lived priming or task pressure without durable learning. The classical observational research tradition investigated these issues by varying the consequences experienced by models and later testing what children could reproduce with incentives. The central lesson of the program is methodological: measures of performed imitation alone do not fully specify acquired information Bandura (1965) Bandura et al. (1963b).
Normative understanding requires an additional inferential step. A child may know how a task is completed yet also believe that members of a group ought to perform it in a particular way. The first attribution concerns a technique or causal procedure; the second assigns a rule to a social activity. Experimental demonstrations of children's protest against a puppet's incorrect game move are especially relevant because the test examines response to another actor rather than simple reproduction by the observer. The same movement can be physically possible in both conditions while its status as a violation depends on the collaboratively established game. Such contrasts make the conventional meaning of an action empirically tractable Rakoczy et al. (2008) Schmidt et al. (2012).
Researchers therefore need to distinguish at least five outcomes: perceptual attention to a demonstration, encoding of its action sequence, capacity to reproduce it, actual performance under current incentives, and treatment of a practice as a socially binding convention. A sixth outcome, successful transmission to a new learner, adds a population-level dimension. These are interrelated, but no single measurement automatically establishes all of them. Multi-stage designs with delayed independent tests, alternative demonstrators, appropriate controls, and changes of social context support more informative causal conclusions. This layered approach also prevents conformity observed in an adult survey from being equated with the developmental origin of that person's normative expectations Legare & Nielsen (2015) Kendal et al. (2018).
Early Experimental Traditions and Social Cognitive Theory
The history of social-learning research includes older work on association, reinforcement, imitation, suggestion, and group influence. A decisive experimental program was developed by Albert Bandura and collaborators during the early 1960s. The 1961 study by Bandura, Ross, and Ross examined whether children exposed to an aggressive adult model later displayed corresponding forms of aggression. Subsequent studies introduced film-mediated modeling and the consequences experienced by observed actors. These studies helped establish that the behavioral repertoires investigated by psychologists could be changed by exposure to social exemplars, extending learning research beyond procedures in which the learner personally receives every consequential reinforcement Bandura et al. (1961) Bandura et al. (1963a).
The observational program developed a useful conceptual separation between what the model demonstrates and what the observer anticipates will follow from reproducing it. Modeling may supply action structure, information about feasibility, and evidence about social approval or sanction. In the 1965 reinforcement-contingency experiment, children observed a filmed model whose aggressive actions were rewarded, punished, or followed by no consequences. Later incentives to reproduce the actions altered measured performance and narrowed previously apparent differences between experimental conditions. The study provides a direct reason to assess learning and performance separately; it does not imply that all observed aggression in everyday settings has an identical causal origin Bandura (1965).
Later social cognitive and cultural-learning traditions widened the question from reproducing isolated movements to acquiring social skills, role models, beliefs, interaction scripts, and expectations about shared practices. Modern research also asks which learners select which models and under what information conditions. The historical studies remain important demonstrations of experimental modeling, while their original samples, settings, behaviors, and ecological transfer require explicit consideration. It would be mistaken to treat the Bobo-doll paradigm as a universal measure of aggression or a complete theory of norm emergence. Its enduring contribution is a controlled demonstration that socially available action information can influence behavior in ways that require experimental analysis of observation, motivation, and response Bandura et al. (1963b) Kendal et al. (2018).
Vicarious Reinforcement and the Meaning of Observed Consequences
Vicarious reinforcement refers to learning or performance changes related to the rewards, punishments, or other consequences observed in another agent's experience. A demonstration contains both an action and an outcome sequence. Learners can use the outcome to estimate whether the action is effective, safe, socially approved, or costly. These possibilities have different psychological implications: the same visible reward may indicate material payoff in one setting and social standing in another. Bandura, Ross, and Ross studied vicarious reinforcement in 1963, while Bandura's later manipulation of model consequences more clearly foregrounded the acquisition–performance distinction Bandura et al. (1963b) Bandura (1965).
Observed consequences may change attention to a model, memory for a behavior, anticipation of sanction, or willingness to act when tested. For causal inference, these paths must be separated by design. An experiment measuring only the frequency of postexposure copying cannot establish whether a rewarded model was encoded more accurately than a punished model; a delayed reproduction test under standardized incentives can address that possibility more directly. Equally, avoidance of copying after punishment can be rational adaptation to expectations about likely social consequences. The behavioral meaning of an observed outcome is therefore conditional on the learner's goals, the credibility of the situation, and the availability of alternative routes to success Kendal et al. (2018) Over & Carpenter (2012).
The role of vicarious outcomes becomes more complex when an observer participates in a community with competing norms. Seeing a transgression rewarded by one subgroup and sanctioned by another can teach the learner both the action and the group-specific contingency governing it. A strong research design would measure the remembered sequence, the learner's expectation of approval, their choice under changed audiences, and the later evaluation of another actor doing the same thing. Such an analysis treats visible consequences as information that is socially interpreted rather than a mechanical guarantee of imitation. Contemporary studies of normative expectations further illustrate why behavior depends on the empirical expectations people form about others' likely conduct Bicchieri & Xiao (2009) Cialdini et al. (1990).
From Individual Modeling to Cultural Inheritance
Social-learning research joins cultural evolution when behaviors and representations are transmitted repeatedly among individuals across time. Boyd and Richerson formalized cultural inheritance in a population perspective, asking how social transmission interacts with individual learning, selection, and variation. Here a behavior's distribution is a joint outcome of learning rules, environmental opportunities, network structure, and the history of contact among learners. The unit of analysis expands from one observer–model encounter to a sequence of transmissions. A population's shared custom may persist even when no single participant consciously planned its long-term preservation Boyd & Richerson (1985) Mesoudi & Whiten (2008).
Formal theories distinguish learning from personal experience and learning from social information, then examine conditions under which each is useful. Rogers demonstrated that the evolutionary consequences of cultural learning cannot be assumed from the mere existence of copying; conditions that make social learning attractive for an individual do not automatically guarantee higher average population performance. Later strategy tournaments investigated the profitability of copying with explicit computational environments. These results demonstrate the importance of context, access to information, and the selective sampling of successful demonstrations rather than a single universal rule that social information always improves behavior Rogers (1988) Rendell et al. (2010).
An account of norm acquisition must explain both the initial appearance of a convention and its persistence across repeated social interactions. Individual encoding produces a repertoire; teaching, copying, correction, reward, and group membership shape which elements reappear; cultural transmission generates distributions that subsequent learners encounter as their social environment. The apparent stability of a practice may therefore be a product of recurrent local learning rather than unchanged mental representations. Carefully designed transmission-chain experiments can examine how variations accumulate, while models of cultural inheritance specify the assumptions required for persistence and turnover. This multilevel connection supplies a framework for understanding social practices without confusing a population pattern with a proven psychological mechanism in each individual Legare (2017) Kendal et al. (2018).
Attention, Retention, and Model Selection
Social information becomes available to a learner only after a demonstrator's activity is perceived and treated as relevant. Attention may follow familiarity, confidence, apparent success, emotional salience, or opportunities for repeated interaction. Retention requires encoding enough of the observed action and context to influence a later encounter. The selection of a demonstrator introduces an additional psychological problem: which people provide reliable information for this particular task? Social-learning strategies research analyzes choices about when to copy, whom to copy, and what information to adopt. The same strategy label can describe different underlying processes, which makes mechanism-specific experiments essential Kendal et al. (2018) Henrich & Gil-White (2001).
A learner may observe several models producing incompatible actions. Choice among them can depend on demonstrated competence, relative frequency, prestige, kinship, or shared identity. Apparent majority copying may reflect counting independent people, repeated exposure, or perceived social affiliation. Experiments can separate these explanations by holding the number of demonstrations constant while varying the number of distinct demonstrators. Such a design has been applied comparatively to human children and great apes. The findings establish selective use of social information in the tested groups while leaving open how motivational and cognitive processes vary across tasks and populations Haun et al. (2012) Over & Carpenter (2012).
Attention and retention can also be measured independently of performance. Eye tracking can document visual allocation during demonstration; surprise tests can assess retained steps; timed transfer tasks can show how a learned action is used later. These measures should be interpreted with care because looking at a hand movement is not equivalent to understanding its purpose, and remembering a sequence is not equivalent to accepting it as a rule. An informative design combines multiple observations, randomizes model conditions, defines preregistered outcomes when appropriate, and examines whether model-selection effects survive changes of audience and reward structure Kendal et al. (2018) Legare & Nielsen (2015).
Imitation, Emulation, and Understanding Goals
Imitation reproduces important features of another agent's action, whereas emulation uses observed information about an outcome or environmental relationship to achieve a result, potentially by different means. The distinction helps researchers ask whether an observer learned the demonstrator's movements, the task's causal structure, or both. Horner and Whiten manipulated the availability of causal information in tool-use problems for chimpanzees and children. Their experiments illustrate that observed behavioral matching has to be interpreted against what learners know about the apparatus and which features of the task are visibly necessary Horner & Whiten (2005).
Human children sometimes reproduce apparently irrelevant steps, a phenomenon termed overimitation. One explanation emphasizes uncertainty about physical causation: perhaps the observer thinks the extra step is necessary. Another emphasizes social or normative interpretation: perhaps the child regards the demonstrated method as the established way to act. Kenward and colleagues tested predictions that distinguish these accounts, providing evidence compatible with normative learning in their experimental conditions. Nielsen and Tomaselli investigated overimitation in populations with markedly different everyday learning environments, offering important cross-cultural comparison without establishing that all populations copy the same way in every setting Kenward et al. (2011) Nielsen & Tomaselli (2010).
Imitation is therefore a flexible response with multiple functions. An individual can imitate to obtain a reward, to communicate affiliation, to express respect, to learn an unfamiliar procedure, or to preserve a conventional ritual whose practical purpose is secondary. The interpretation depends on experimental manipulations and the learner's stated or inferred goal. A robust test of norm acquisition includes a violation probe: will the observer protest when a third party uses a different method, even if that method reaches the same goal? A robust test of instrumental learning instead asks whether the learner selects an efficient means when the causal structure changes Legare & Nielsen (2015) Over & Carpenter (2012).
Outcome Learning, Prediction, and Social Evaluation
Observing others creates opportunities for prediction about consequences. An observer may learn that a behavior produces an object, receives praise, triggers exclusion, or changes another person's response. Such information can update expected value and guide selection of later action. In historical modeling experiments, consequences administered to the demonstrator affected children's postexposure behavior, but the relation between learned representation and expressed imitation required subsequent incentive probes. The appropriate causal question is therefore which learning variable the experimental outcome identifies, rather than whether the observed consequence has a generically positive or negative meaning Bandura (1965) Bandura et al. (1963b).
Neuroscientific investigations provide a complementary level of analysis. Klucharev and colleagues studied changes in attractiveness judgments after participants encountered group opinions and reported neural signals associated with disagreement and later adjustment. Their findings connect some forms of social conformity to mechanisms resembling reinforcement-learning prediction errors. The measured response concerns a particular judgment paradigm and cannot, on its own, demonstrate the neural origin of every childhood convention or durable cultural tradition. Social evaluation, explicit expectation, and embodied action learning can recruit overlapping resources while remaining distinct explanatory targets Klucharev et al. (2009) Bicchieri & Xiao (2009).
Experiments should compare the effects of observed outcome, perceived authority, stated rule, and majority consensus rather than treating all social information as one stimulus. For example, a high-status demonstrator who fails at a task may still be copied by a learner seeking membership, while a competent unfamiliar demonstrator may be preferred by a learner seeking accuracy. Factorial manipulation of competence and affiliation could identify their relative contributions under different goals. Delayed behavioral assessments would then determine whether the intervention changes a retained strategy, an immediate expression, or an expectation of social reward. Such designs operationalize the conditional nature of model use emphasized by social-learning strategies theory Henrich & Gil-White (2001) Kendal et al. (2018).
Teaching, Scaffolding, and Reciprocal Social Interaction
Many important practices are acquired through interaction in which a more experienced participant actively directs the learner's attention, models a sequence, corrects errors, or provides opportunities to participate. Teaching changes what information is displayed and when feedback becomes available. The distinction between passive exposure and directed instruction is especially important for social norms: an adult can show a procedure and also indicate that the demonstrated method is customary or expected. Developmental research on cumulative cultural learning emphasizes the coevolutionary significance of teaching practices and children's flexible learning strategies Legare (2017).
Scaffolding is a useful descriptive concept for temporally structured support that allows a learner to perform an activity beyond what a single independent encounter would permit. Its study requires observed sequences of assistance, independent performance after support is reduced, and retention in new settings. Joint participation may convey a role structure—who begins, who waits, what counts as completion—that a solitary action demonstration would leave ambiguous. Shared intentionality theory offers a broader account of how children participate in common practices involving mutually recognized intentions and conventions. It is a theoretical interpretation requiring comparison with alternative explanations and direct behavioral tests Tomasello & Rakoczy (2003) Rakoczy et al. (2008).
Social transmission is reciprocal when learners subsequently become demonstrators, interpreters, or enforcers for others. An initially taught action can be reformulated in explanation, simplified for a novice, or augmented with a justification. Transmission-chain designs reveal whether such transformations improve stability, introduce innovation, or change what participants believe is mandatory. The learner's role may alter what gets transmitted: an individual who can reproduce a practice may be unable to explain its conditions of use, while an explicit teacher may articulate a rule without always performing it. Examining these differences connects laboratory research with schooling, family socialization, apprenticeships, and community practice Mesoudi & Whiten (2008) Legare & Nielsen (2015).
The Developmental Emergence of Conventional Rules
Norm acquisition becomes visible in early childhood when a child treats an observed practice as a conventional way of doing something. Developmental studies use simple games with unfamiliar rules because the researcher can introduce a new convention and distinguish its correct and incorrect forms. In experiments by Rakoczy, Warneken, and Tomasello, children observed a game, played it, and then watched a puppet perform an act that violated the established procedure. Children aged two and three responded more normatively in relevant violation conditions than in controls, with more explicit responses among the older children. The evidence concerns early understanding of conventional activities rather than a complete mature moral system Rakoczy et al. (2008).
Learning a rule requires some representation of its domain. A gesture may count as incorrect in one jointly established game while remaining acceptable in another, and observers must track this contextual assignment to protest coherently. Accordingly, developmental tests gain force when the physical action stays constant while the conventional context changes. The response to a violation can be recorded as stopping, correcting, teaching, or explicit verbal protest. These behaviors provide convergent but nonidentical indications of normative understanding. Language ability and willingness to confront a puppet may also affect the measure, so developmental comparisons should avoid equating low protest with absence of relevant knowledge Rakoczy et al. (2008) Schmidt et al. (2012).
Conventional rule acquisition is inseparable from the child's expanding participation in shared activities. Learning how a joint activity is organized includes expectations about others' roles, permitted alternatives, and how deviations should be interpreted. The account developed by Tomasello and Rakoczy links early social understanding with shared intentionality and the conventional character of group practices. The framework identifies research questions about coordinated goals and common knowledge, while experiments must still establish which specific information a child used. Direct manipulation of joint participation, verbal instruction, and repeated demonstration can test the mechanisms proposed in this developmental account Tomasello & Rakoczy (2003) Legare (2017).
Inferring Norms from Limited Demonstration
Humans sometimes infer that a particular behavior is a rule after seeing very little evidence. Schmidt, Butler, Heinz, and Tomasello experimentally investigated whether three-year-old children could infer a social norm from a single adult action in a context where explicit normative instruction was absent. The children in their reported study sometimes treated an observed act as establishing a proper way of behaving and reacted when another agent deviated from it. This finding is important because it shows that prescriptive interpretation can arise without an explicitly uttered command, although the task's communicative circumstances remain part of the causal setting Schmidt et al. (2016).
Inferring a rule from one demonstration could be aided by children's prior experience of adult instruction, assumptions that demonstrations are purposeful, or expectations about conventional games. An experiment showing spontaneous protest does not uniquely determine which of these inferences was involved. Researchers can vary whether an adult acts intentionally or accidentally, whether observers share the demonstrator's group identity, and whether the action has a transparent instrumental purpose. A later teaching test could show whether the child considers the action transferable to another newcomer. These contrasts help specify when sparse evidence is treated as representative of an established convention and when it is interpreted as an idiosyncratic choice Schmidt et al. (2016) Over & Carpenter (2012).
Single-event norm inference also poses a general problem for social epistemology. Learners must extrapolate from observations that are usually selective and incomplete. If an action appears repeatedly, that frequency may supply information about what most participants do; if it is performed once by a trusted teacher, social status may supply another cue. The two sources can make competing predictions. Experimental designs that cross number of models, model competence, and presence of directive language would help separate frequency-based induction from authority-based convention learning. Population-level regularity alone cannot establish that learners believe the practice is obligatory Haun et al. (2012) Bicchieri & Xiao (2009).
Protest, Correction, and Norm Enforcement
Norm enforcement is behavior through which an observer objects to, corrects, or sanctions another person's perceived violation of a rule. In developmental research it is especially informative because the learner applies a convention to a third party instead of simply copying the original demonstration. Rakoczy and colleagues observed normative responses during rule-game violations, while Schmidt and colleagues examined whether children respond selectively when the violator belongs to a relevant group. These methods connect acquired rule representation to outward intervention but do not establish that every protest arises from an abstract moral commitment Rakoczy et al. (2008) Schmidt et al. (2012).
Enforcement can serve different ends. A participant may correct another actor to maintain coordination, restore a valued convention, signal group identity, or avoid downstream confusion. A child might protest more often when an ingroup partner violates a rule because that partner is expected to know it, or because the interaction threatens a common activity. Experimental comparisons should examine the learner's expectations about what an outsider could reasonably know, not solely the raw number of protests. This is one reason group affiliation is a theoretically meaningful manipulation and a potential source of heterogeneity in norm-enforcement findings Schmidt et al. (2012) Tomasello & Rakoczy (2003).
An advanced account of norm acquisition distinguishes recognition of a violation, motivation to enforce, and the ability to communicate a correction. Those capacities may develop at different rates and vary with language, interpersonal risk, and familiarity. A child who silently predicts that an action is inappropriate has not necessarily acquired a weaker rule than one who loudly corrects a puppet. Researchers can combine third-party judgment, helping behavior, spontaneous intervention, and verbal explanation to reduce dependence on any one expression channel. This measurement principle extends to adult norm learning and cross-cultural research, where public sanctions and private evaluations can also diverge Rakoczy et al. (2008) Bicchieri & Xiao (2009).
Peer Models, Adolescence, and Social Scripts
Peer groups supply recurring models for speech, dress, cooperation, risk taking, and the practical organization of everyday relationships. A social script is a learned expectation about the usual sequence of actions and responses in a recognizable interaction. A learner can acquire such scripts through repeated observation without being formally instructed. The scientific problem is to determine what exposure teaches about possibilities, what peer approval teaches about incentives, and what group membership teaches about expected conduct. Classical modeling and social-motivation research provide foundations for analyzing these pathways, but direct claims about adolescence require designs specific to adolescents and their contexts Bandura (1965) Over & Carpenter (2012).
Adolescent social learning should be analyzed through reciprocal peer selection and influence. People often choose associates who already resemble them, while interaction may further change their behavior. A cross-sectional correlation between a young person's conduct and peer norms cannot determine the direction of causation. Longitudinal network designs, randomized peer exposure when ethical, or natural changes in classroom organization can improve identification. The researcher should also distinguish perceived popularity, actual behavioral prevalence, and expectations about approval, because they can have different effects on whether a practice is adopted, concealed, or abandoned Bicchieri & Xiao (2009) Centola & Macy (2007).
Social scripts can be adaptive for rapid coordination and harmful when they reproduce coercive or exclusionary practices. The same general acquisition machinery can transmit professional courtesy, assistance routines, bullying conventions, or unsafe risk norms. Explaining a script's origin through social learning does not settle its ethical value or its prevalence in a particular population. Interventions can modify who demonstrates behavior, how credible alternatives are made visible, and whether opportunities for independently rewarding conduct are available. The empirical standard is a demonstrable change in sustained behavior with appropriate controls, rather than an immediate increase in declared awareness of desirable norms Cialdini et al. (1990) Centola (2010).
Norm Learning Across the Life Course
Norm acquisition continues after early childhood when individuals enter new institutional, occupational, relational, or technological environments. Adults learn practical rules by watching colleagues, receiving correction, and discovering which conduct facilitates cooperation. The mechanisms of model attention, observed consequences, and social identification remain analytically relevant, although their relative importance can change with existing knowledge, role obligations, and opportunities for independent testing. Adult social learning often involves revising a previous convention rather than forming a representation for the first time. A longitudinal approach should therefore measure both acquisition of the incoming norm and the persistence of the earlier one Kendal et al. (2018) Boyd & Richerson (1985).
Age comparisons can be misleading when participants differ in exposure history. A younger learner may copy an unfamiliar action exactly because its purpose is unknown, while an experienced learner may omit arbitrary details because they can reconstruct the causal goal. Legare and Nielsen's distinction between instrumental and conventional learning clarifies why imitation can remain highly faithful in one domain while innovation increases in another. A useful life-course study would hold the task constant, vary its explicitly conventional versus instrumental framing, and estimate how learning strategies differ with age and prior practice Legare & Nielsen (2015) Legare (2017).
Transitions into a new group expose the learner to competing expectations. A migration context, a new workplace, or a newly joined online community may demand switching between rule sets, interpreting conflicting demonstrations, and deciding when a local practice is binding. Such circumstances can test whether the learner relies on repeated majority behavior, a prestigious representative, a clear instruction, or direct payoff. Cross-cultural and intergenerational designs are particularly valuable because they reveal which pathways generalize and which depend on established educational and socialization arrangements. Claims about universal stages of norm acquisition require populations and measures broad enough to support that inference Nielsen & Tomaselli (2010) Blake et al. (2015).
Majority Influence and Frequency-Dependent Social Learning
Frequency-dependent social learning describes strategies in which the distribution of demonstrated behavior affects what the observer adopts. Conformist transmission refers more specifically to disproportionate adoption of the most common variant under a defined comparison. Henrich and Boyd used formal evolutionary models to investigate conditions favoring such transmission and its consequences for cultural variation. These models clarify what assumptions are required for local traditions to persist; they do not themselves measure the cognitive algorithm of a particular child. In experiments, researchers must distinguish the proportion of individuals demonstrating a behavior from the sheer number of demonstrations the observer receives Henrich & Boyd (1998) Haun et al. (2012).
Haun, Rekers, and Tomasello separated the number of demonstrators from the number of demonstrations in a comparative experiment: one action was performed by three different individuals, while a competing action was performed repeatedly by one individual. Their findings supported a majority-biased response in human children and chimpanzees under the tested conditions, with a different result in orangutans. This specificity matters: a species comparison within one apparatus is evidence about a strategy in that setting, while claims about a species' comprehensive cultural abilities require broader sampling and independent tasks Haun et al. (2012).
Population frequencies can also become cues to safety or social belonging. A novice may copy a common practice because many people independently found it effective, because it reduces social friction, or because alternative choices may be punished. These mechanisms can yield similar aggregate conformity but diverge when incentives or social visibility are manipulated. Kendal and colleagues emphasize that observed social-learning strategies are flexible and that population patterns need not map one-to-one onto individual decision heuristics. The next scientific step is to test the mechanism generating frequency-sensitive acquisition with controlled counterfactuals Kendal et al. (2018) Bicchieri & Xiao (2009).
Prestige, Competence, and the Credibility of Models
Prestige-biased learning prioritizes information from agents perceived as admired, respected, or socially deferred to. Henrich and Gil-White proposed an evolutionary account in which freely conferred deference can help learners identify potentially valuable cultural information. The theory raises a useful distinction between influence based on coercive power and influence based on perceived informational competence. In real communities these signals may overlap, and a learner may choose an admired model for affiliative rather than instrumental reasons. Evaluating the theory thus requires measuring what status cue was observed, what the learner believed about expertise, and which behavior was actually acquired Henrich & Gil-White (2001).
Expertise is often domain-specific. Someone who is an excellent craft demonstrator may have little reliable knowledge about interpersonal fairness or unfamiliar technology, but learners may transfer reputation across domains. A well-designed experiment should assign observable competence in one task, manipulate prestige separately, and later measure copying across matched and unmatched domains. Demonstrator identity should also be distinguished from the content of the demonstrated action. Without this separation, a behavioral association with popularity cannot establish prestige-biased information acquisition as the causal mechanism Kendal et al. (2018) Henrich & Gil-White (2001).
Model credibility is dynamic rather than fixed. Learners may initially rely on deference cues and revise those judgments after watching repeated success or failure. A cultural-learning explanation should permit both social selection and individual evaluation of what was copied. This leads to questions about mechanisms that preserve valid expertise and pathways that amplify inaccurate traditions when high-status demonstrators are imitated outside their competence. Contemporary information environments make the distinction especially salient because platform prominence can appear to signal authority while being generated by attention-ranking processes. Such extensions are plausible research hypotheses and need direct experiments on model interpretation Rendell et al. (2010) Centola (2010).
Animal Traditions and Comparative Transmission
Comparative psychology investigates whether animals acquire behavioral variants socially and whether groups sustain traditions over repeated encounters. Whiten, Horner, and de Waal experimentally introduced different tool-use methods into chimpanzee groups and observed their spread within groups. The study supplied controlled evidence of cultural transmission in a setting where natural field observations alone could not fully identify how traditions had originated. The strongest warranted conclusion concerns socially transmitted group-specific practices and local convergence; it does not by itself show that chimpanzees conceptualize their practices as prescriptive obligations in the human sense Whiten et al. (2005).
Comparative tasks can identify action copying, outcome emulation, and sensitivity to numbers of demonstrators. Horner and Whiten's manipulations of causal visibility addressed imitation–emulation switching, while Haun and collaborators investigated whether majority representation affected learning among chimpanzees, orangutans, and children. Together, such studies show that specific social-learning strategies can be tested across species, but the comparability of apparatus, prior experience, social ecology, and motivation requires careful control. What counts as an easy action for one species may carry a different motor or perceptual cost for another Horner & Whiten (2005) Haun et al. (2012).
Human normative learning involves an additional inferential issue: observers may treat a demonstrated sequence as defining how group members should act, and they may correct deviations by third parties. Comparable evidence in animals would require operational indicators beyond stable local copying, including context-sensitive reactions to violations and experimental exclusion of explanations based on expected reward or disruption. The human developmental findings are therefore a useful hypothesis-generating reference, not a license to reclassify every animal tradition as norm enforcement. The comparative field becomes stronger by identifying separate continuities in learning mechanism, population transmission, and prescriptive understanding Rakoczy et al. (2008) Whiten et al. (2005).
Cumulative Cultural Learning, Ritual, and Innovation
Cumulative cultural learning occurs when socially transmitted practices are retained and modified across successive learners, producing outcomes that depend on contributions accumulated over time. This process combines fidelity, selective preservation, innovation, teaching, and opportunities for communication. Legare emphasized that acquisition strategies in childhood vary with the kind of practice being learned, while Mesoudi and Whiten reviewed experimental methods for studying cultural transmission. A single act of copying demonstrates access to an action model; a transmission chain adds evidence about how representations survive or change across generations of learners Legare (2017) Mesoudi & Whiten (2008).
Instrumental tasks invite observers to evaluate whether steps contribute to a desired result. Conventional or ritualized tasks may instead emphasize preserving a socially recognized form. Legare and Nielsen synthesized evidence that imitation and innovation are differently combined in these circumstances: experience can promote flexible alteration of instrumental methods while conventional learning may preserve high-fidelity performance. Overimitation experiments add detail by showing that seemingly unnecessary steps can acquire social significance. The interpretation of such tasks depends on whether participants were invited to maximize efficiency, reproduce a demonstration, participate in a shared ritual, or teach a newcomer Legare & Nielsen (2015) Kenward et al. (2011).
A complete theory must explain both the fidelity of cultural reproduction and the arrival of new variants. Uncritical copying can conserve complex sequences whose functions novices cannot independently reconstruct, while selective innovation can adapt practices to new conditions. Excessive fidelity may perpetuate ineffective details; excessive alteration may dismantle a coordination convention. Experiments can track each step across successive participants, record the explanations they provide, and compare information loss with behavioral performance. Cultural-evolution models supply predictions about long-run distributions, while developmental psychology supplies mechanisms of learning and teaching that require independent measurement Boyd & Richerson (1985) Mesoudi & Whiten (2008).
Cultural Diversity, Socialization, and Generalizability
Learning environments vary across societies in who teaches, which demonstrations are considered informative, how explicit rules are communicated, and when children are expected to observe without intervention. A cross-cultural psychology of social learning asks how these arrangements affect the development of imitation, innovation, and normative enforcement. Nielsen and Tomaselli compared overimitation in settings differing in social organization, illustrating why laboratory results from one urban population cannot stand as a universal standard. Legare's synthesis likewise treats variation in caregiver practice and learning strategy as part of the evidence necessary to explain cumulative cultural learning Nielsen & Tomaselli (2010) Legare (2017).
Cross-cultural evidence on the development of fairness further cautions against assuming that every apparent norm is acquired at one invariant age. Blake and collaborators experimentally studied children's responses to unequal allocation in seven societies and reported different developmental patterns for disadvantageous and advantageous inequity aversion. These outcomes concern fairness judgments and behavior, not direct observation of how each particular local fairness norm was taught. They nevertheless identify important variation that any broader model of norm acquisition must eventually explain, potentially through socialization, ecology, institutions, and individual development Blake et al. (2015).
Researchers must distinguish cultural differences in the mechanism of learning from differences in what is demonstrated and rewarded. Two populations might share the same sensitivity to competent models yet acquire different conventions because their social environments supply different practices. Conversely, superficially similar customs might be maintained by different pathways. Evidence requires comparable measures, local linguistic validation, sampling beyond schools and universities, and collaborative interpretation with researchers familiar with each context. Broad claims of universality become defensible when a phenomenon appears across diverse ecologies under tests that make equivalent psychological demands Nielsen & Tomaselli (2010) Kendal et al. (2018).
Experimental Paradigms for Observational Learning
Controlled modeling experiments typically manipulate what demonstrators do, which outcomes follow their actions, and which opportunities observers later receive. The dependent measures may be matching acts, sequence fidelity, latency, spontaneous use in a new context, or transfer to another learner. Bandura and collaborators' early studies remain illustrative because model exposure was experimentally manipulated and behavior was subsequently measured. The 1965 reinforcement-contingency paradigm adds the important step of assessing performance under new incentives. Results should be interpreted in relation to the demonstrated action, measurement window, social meaning of the model, and controls for prior experience Bandura et al. (1961) Bandura (1965).
Norm-learning experiments often use unfamiliar games to create a novel rule with a known reference behavior. The experimenter demonstrates a convention, a third actor violates it, and the learner's response indicates whether the action has acquired prescriptive significance. Suitable controls preserve motor movements while changing their status as rule violations. Studies of spontaneous norm inference introduce a demonstration without explicit instruction and examine whether learners nevertheless correct a later deviation. Each paradigm offers stronger causal evidence than an uncontrolled report that a practice was acquired socially, but each also introduces demand characteristics and assumptions about children's understanding of staged interactions Rakoczy et al. (2008) Schmidt et al. (2016).
An experimental program benefits from separating exposure, acquisition, and follow-up phases. Researchers should specify random assignment, possible social contamination across conditions, coding of behavioral outcomes, blind observation when feasible, and opportunities for delayed assessment. High-fidelity copying can be measured as preservation of demonstrated steps, yet coding must distinguish indispensable and arbitrary components. A rule-judgment task should separately assess correctness, group generalization, and contexts in which deviation is permitted. This combination improves construct validity by asking whether an observed difference is a learned technique, a learned expectation, or a motivation to display agreement Horner & Whiten (2005) Kenward et al. (2011).
Transmission Chains, Diffusion, and Social Networks
Transmission-chain experiments study the transformation of information as it passes from person to person. Researchers can seed a story, artifact procedure, or novel convention with the first participant, then observe each subsequent learner's reproduction or modification. This design reveals fidelity, systematic selection, innovation, and information loss across repeated encounters. The interpretation depends on whether participants can interact freely, receive corrective feedback, or observe only the preceding participant. Mesoudi and Whiten surveyed such experimental approaches as tools for relating microlevel social learning to the evolution of population traditions Mesoudi & Whiten (2008).
Networks introduce a further structure because transmission follows relationships rather than a simple line. Centola and Macy theorized that practices requiring social reinforcement from multiple contacts can spread differently from simple information. Centola later ran an online social-network experiment designed to examine the spread of health-related behavior under different network structures. These sources connect diffusion to clustered contact and reinforcement, but the degree to which an observed adoption reflects internalized norm acquisition remains an empirical question. A network may facilitate public action while leaving private endorsement unchanged; the dependent variables must be specified accordingly Centola & Macy (2007) Centola (2010).
Large-scale digital experiments can test influence with real behavior, while also raising difficult privacy and external-validity questions. Bond and colleagues reported a randomized Facebook political-mobilization experiment involving approximately 61 million users, studying effects of social information and subsequent activity. Its scale and design provide evidence that mediated social signals can affect behavior and diffuse through social ties. Yet political mobilization after a message is not equivalent to the acquisition of a general prescriptive rule. For research on norms, investigators should add assessments of persistence, transfer, reference-group identity, and perceived obligation, rather than equating every network effect with cultural learning Bond et al. (2012) Bicchieri & Xiao (2009).
Measurement, Operational Definitions, and Construct Validity
Social-learning constructs require indicators appropriate to their conceptual definitions. Imitation fidelity may be scored from recorded action sequences, emulation from successful goal attainment by an alternative method, and norm inference from spontaneous correction after a convention is introduced. Majority bias requires a design in which multiple individual demonstrators can be distinguished from multiple repetitions by one demonstrator. Prestige bias requires independently measured prestige or deference rather than an assumed relationship between popularity and expertise. These operational distinctions prevent an identical observed response from being assigned incompatible mechanisms without evidence Haun et al. (2012) Henrich & Gil-White (2001).
Reliability involves consistent coding and measurement across observers, trials, and contexts. Validity concerns whether the measure actually represents learning, social motivation, normative interpretation, or a different construct. A child might remain silent during a puppet's violation because of shyness, limited expressive language, or uncertainty about the experimenter's expectations. A low protest score should therefore be interpreted together with comprehension checks and less confrontational rule-judgment measures. Similarly, adult reports of customary behavior can reflect impression management or uncertainty about what others do. Triangulating behavior, expectations, and subsequent transfer improves interpretation Rakoczy et al. (2008) Bicchieri & Xiao (2009).
Experiments should disclose task instructions, stimulus materials, exclusion criteria, coder agreement, assignment procedures, and plans for analyzing repeated measures. For cultural comparisons, equivalent translation is only the beginning of comparability; the task must also be familiar enough to support meaningful participation and unfamiliar enough to measure new learning. For computational approaches, model fit should be evaluated against behavior not used to construct the model, and alternative parameterizations should be considered. The scientific value of a measure depends on the inference it permits, not the familiarity of its label Kendal et al. (2018) Mesoudi & Whiten (2008).
Causal Identification, Replication, and Evidence Quality
Claims about social learning are strongest when exposure to a social model is manipulated while other information sources are held constant. Observational studies are still valuable for describing natural environments, but people often select their friends, schools, occupations, and information networks. Similarity among group members can consequently arise from shared background and homophily as well as from influence. Experiments can break part of this confounding through randomized assignment of demonstrators or norm information; longitudinal studies can test temporal sequences; and network designs can analyze exposure patterns while stating the assumptions required for causal interpretation Centola (2010) Bond et al. (2012).
Independent replication should test whether an effect persists across materials, ages, cultural contexts, and laboratories. Classical modeling studies established influential paradigms, but transferring their conclusions to modern media or adult institutions requires new evidence, ideally with multiple outcomes and preregistered analysis plans. Mechanisms of normative protest may also be sensitive to how a puppet's error is introduced, how adults respond, and what children expect a participant to know. Replication across changed procedures can identify a robust construct or clarify the limits of a familiar result. Effect sizes, uncertainty intervals, and transparent reporting help readers judge practical importance when numerical results are available Bandura (1965) Rakoczy et al. (2008).
The body of evidence is heterogeneous: randomized human experiments, carefully controlled comparisons with animals, computational tournaments, developmental observations, and cultural-evolution models answer distinct questions. Formal simulations test consequences of assumptions rather than establish how real learners behave. A field experiment may be ecologically rich while leaving the mechanism uncertain. A laboratory experiment may identify a mechanism narrowly while leaving long-term transfer unresolved. Critical synthesis should therefore compare the causal target, sampling frame, measures, and alternative explanations before declaring findings convergent. The present monograph is a critical narrative synthesis and makes no claim to a protocol-registered systematic review Rendell et al. (2010) Kendal et al. (2018).
Theoretical Tensions and Unresolved Mechanisms
Several theories can explain similar copying behavior. An instrumental account emphasizes estimated outcomes; a social-motivational account emphasizes affiliation; a cultural-learning account emphasizes transmission rules; and a normative-cognition account emphasizes representation of socially binding conventions. A child copying an inefficient procedure might be uncertain about the causal apparatus, preserving a ritual convention, seeking approval, or some combination. These possibilities generate different predictions when the observer is tested privately, when the outcome becomes transparent, or when a third party deviates. The scientific challenge is to design experiments that discriminate rather than merely redescribe observed fidelity Horner & Whiten (2005) Kenward et al. (2011) Over & Carpenter (2012).
A second tension concerns the relation between population-level conformist patterns and individual cognitive heuristics. Formal models often use simplified transmission rules to predict stable cultural variation, while developmental and comparative experiments reveal flexible choices among demonstrators. Majority-biased behavior can follow several processes, including repeated social reinforcement or deference to individuals perceived as representative. Kendal and colleagues explicitly caution that learning strategies may be deployed together and that inferred population patterns need not identify a unique mental rule. Research gains precision when the same study connects observable action, stated expectations, and independently estimated social-network exposure Henrich & Boyd (1998) Kendal et al. (2018).
A third issue concerns whether normative understanding is continuous with other forms of learning or depends on additional capacities for shared intentionality, convention, and entitlement. Developmental protest and rule-inference experiments document important phenomena, while comparative studies reveal animal social transmission without demonstrating identical norm concepts. Cultural-learning perspectives propose that psychological capacities themselves can be shaped through developmental interaction with socially transmitted practices. These accounts differ in scope and explanatory commitments. Productive research can test their predictions through cross-context transfer, developmental trajectories, comparative designs, and manipulations of communicative intention Tomasello & Rakoczy (2003) Heyes (2019) Whiten et al. (2005).
Recent Experimental Advances in Norm Learning and Emergence
Contemporary research has moved beyond measuring immediate imitation to isolating how prescriptions are inferred, transmitted, and transformed. Malle and Chi's 2024 review organizes norm representations in terms of properties that include context specificity, deontic force, and perceived prevalence, and examines learning by observation, communication, modeling, teaching, and feedback. Its analysis shows why a norm's being widespread, felt as obligatory, or applicable to a particular situation can change through different learning routes. These dimensions make stronger experimental predictions than a single score of agreement with a rule: one intervention could alter beliefs about prevalence while another changes perceived obligation. The review supplies an interpretive framework, while direct evidence about transitions between these properties depends on the designs of individual studies Malle & Chi (2024).
A particularly informative 2025 experiment addresses how a prescription can emerge through successive transmissions rather than merely be copied from an existing authority. Partington, Kamtekar, and Nichols examined iterated teaching about initially inadvisable actions across five studies involving 3,688 participants. As information passed between generations of simulated teachers and learners, later participants increasingly treated certain actions as impermissible and punishable. The authors interpret the pattern in terms of a conjunction between teachers' tendency to communicate bare directives and learners' tendency to infer normative force from these directives. This provides experimental evidence for one route by which transmission can strengthen a rule's prescriptive character. It does not imply that every real-world prohibition historically originated through this mechanism or that its institutional durability has already been demonstrated Partington et al. (2025).
Recent developmental work has also clarified the limits of transferring laboratory results to ordinary social life. Bach and Pauen's study of 160 German children aged four to five contrasted home versus laboratory settings and normative versus nonnormative language in an overimitation task. Their reported interaction indicated that normative wording increased imitation in the laboratory but not in the home setting. This provides a concrete boundary condition for interpreting classroom-like demonstration tasks and warns against inferring setting-invariant norm learning from one laboratory procedure. Gönül and Clément's registered report used affective reactions and dominance cues in staged rule-learning scenarios involving toddlers and preschoolers; their results were differentiated by age and trial, rather than uniformly favoring an authority-driven explanation. Together these studies highlight the need to manipulate context and separate expectations from visible copying Bach & Pauen (2026) Gönül & Clément (2025).
Other recent studies examine what learners infer from a social environment and how normative information modifies their psychological processing. Schroeder, Darling, and Blake found that children's inferred prosocial norms in a novel social setting predicted context-specific prosocial choices following brief exposure, while generalization to a distinct helping task showed a different pattern. Chai and colleagues experimentally investigated majority influence on children's sharing behavior, connecting the number of observed generous models with a specific prosocial outcome. Germar, Albrecht, and Mojzisch measured steady-state visual evoked potentials after social norm learning and reported feature-based attentional alignment with norm-congruent information. Each study identifies a distinct endpoint—local giving, majority-sensitive sharing, or visual attention—rather than an interchangeable measure of universal norm endorsement Schroeder et al. (2025) Chai et al. (2024) Germar et al. (2023).
Norm acquisition also interacts with emotional interpretation. Riggs and Fast studied 148 children aged four to seven and found that participants attributed relatively greater happiness to characters following a conventional norm, including situations where adherence conflicted with desire or freedom of choice. The evidence concerns children's predictions about another person's emotional experience, not a direct measurement of the learner's own joy or a universal emotional reward from compliance. Read alongside experimental work on norm inference, social correction, and observed consequences, these findings open a research program that jointly measures rule understanding, affective expectations, publicly expressed behavior, and transfer across groups and settings. Their methodological value lies in expanding the set of mechanisms and outcomes that norm-learning theories must explain Riggs & Fast (2025) Schmidt et al. (2016) Malle & Chi (2024).
Learning Cooperative and Harmful Practices
Social learning transmits practices with widely different consequences for the people involved. Learners can acquire cooperative routines, methods of mutual aid, equitable allocation, exclusionary conventions, or aggressive scripts by observing people who are salient within their social environments. Modeling theory describes a route by which available examples can affect later action, and developmental work shows that children may also interpret the manner of performance as conventional. The specific content of a transmitted practice must be evaluated through independent evidence of its outcomes. A general learning mechanism can therefore sustain useful coordination or the reproduction of practices that cause harm, depending on the demonstrations, incentives, and social relationships involved Bandura et al. (1961) Rakoczy et al. (2008).
Cooperation can be facilitated when socially learned expectations make others' behavior more predictable. If participants know how to alternate turns, share scarce resources, or signal responsibility, they can coordinate without reinventing procedures in every interaction. Yet the behavioral expression of fairness must be interpreted within a particular society and experimental setting. Blake and colleagues demonstrated important developmental and cross-societal variation in children's responses to unequal allocations. Their results provide comparative evidence about fairness behavior but do not directly identify how each child learned the relevant local conventions. A causal explanation of norm acquisition would require observing or manipulating the socialization history that precedes those judgments Blake et al. (2015) Legare (2017).
Harmful practices can be socially reinforced when observers see people rewarded for intimidation, ostracism, or coercion. Experimental modeling identifies possible acquisition pathways, while naturalistic longitudinal work is needed to estimate the contribution of peer selection, shared institutions, and repeated exposure to long-term conduct. A prevention intervention can offer credible alternative models, make sanctions and social consequences legible, and provide opportunities to practice constructive behavior. To establish its effectiveness, investigators should measure independent behavior beyond the training setting, track durability, and compare it with a suitable control condition. Immediate endorsement of a desirable value is a distinct endpoint from sustained behavioral change Bandura (1965) Bicchieri & Xiao (2009) Centola (2010).
Socialization in Education, Organizations, and Communities
Schools, workplaces, and communities are environments where newcomers acquire rules that are often communicated through repeated practice. Students learn participation conventions while watching teachers and classmates; apprentices learn safety routines from experienced colleagues; community members learn procedures for exchange and collective action. The mechanisms include attention to demonstrators, imitation of action sequences, correction of deviations, and selection among apparently credible models. Educational interventions can use these processes deliberately through visible demonstrations, guided rehearsal, and feedback. Their success should be judged by whether participants acquire and retain the target behavior in settings where supervision and immediate rewards differ from those during instruction Legare (2017) Kendal et al. (2018).
Instructional contexts vary in whether fidelity or adaptive modification is the desired outcome. An unfamiliar laboratory-safety procedure may require precise reproduction because omitted steps alter risk, whereas a creative task may benefit from understanding the goal and selecting an efficient alternative. Research distinguishing imitation from emulation can guide the design of demonstrations so that learners understand which components are causally essential and which are conventional preferences. The relation also runs in the opposite direction: a stable convention can help people coordinate even when several physically equivalent methods would work. Effective instructional assessment must therefore specify which dimension is being taught and avoid rewarding arbitrary fidelity when transfer is the real educational objective Horner & Whiten (2005) Legare & Nielsen (2015).
Organizations often contain competing local norms. An official handbook may prescribe one procedure while repeated examples from respected colleagues convey another. The learner must weigh written rules, observed practice, demonstrator competence, and possible consequences of deviation. A field study can examine how newcomers update their expectations across encounters and whether they later reproduce the practice without prompting. Network structure also matters: repeated exposure from several mutually connected colleagues may strengthen adoption of practices that require coordination. Theoretical and experimental work on complex contagion provides a basis for testing such effects, but claims about internalized workplace norms require measures beyond membership in the network Centola & Macy (2007) Centola (2010).
Digital Networks and Mediated Models
Digital environments multiply the number of models a learner can encounter and change how those models are selected. A platform may present highly visible actions, reactions, and apparent indicators of approval without revealing how the sample was produced. A popular behavior may therefore appear to be a community norm because it is repeatedly displayed, even if its visibility reflects ranking or recommendation procedures. This is an empirically testable problem of exposure and social inference. The distinction between majority information, prestige, and the total number of repetitions becomes especially important where users cannot easily identify whether many independent people or one frequently promoted account generated the observed pattern Haun et al. (2012) Henrich & Gil-White (2001).
Experimental studies of online diffusion demonstrate that behavior can change following mediated social information. Centola's network experiments evaluated how the pattern of social ties affects adoption, while Bond and colleagues used a randomized intervention involving a very large social-network population to study political mobilization and social influence. These studies establish that digital social environments can create measurable behavioral effects under defined conditions. They do not prove that every resulting click, vote, or registration reflects the acquisition of a durable norm. A direct study of digital norm learning would incorporate delayed probes, independent judgments about appropriateness, and transfer to contexts beyond the platform Centola (2010) Bond et al. (2012).
Digital learning research must also attend to algorithmic selection and the diversity of available examples. If a recommender repeatedly amplifies a narrow set of behaviors, an observer may overestimate their prevalence or social legitimacy. Investigators should manipulate model diversity, displayed popularity, and explicit descriptions of how examples were selected, then assess both beliefs and subsequent conduct. The analysis needs to separate misinformation about a practice from the social process by which an observer decides that the practice is customary. These tasks extend established theories of social learning to mediated environments while leaving open empirical questions about persistence, exposure history, and the generalization of laboratory demonstrations Cialdini et al. (1990) Kendal et al. (2018).
Human–Artificial Demonstration and the Artificial Era
The Artificial Era introduces nonbiological systems that can supply demonstrations, instructions, feedback, conversational examples, and apparent social evaluations. For a human learner, an artificial model can function as a source of task information, a visible reference for behavioral scripts, or a participant in an interaction that resembles social tutoring. The empirical question concerns which features of the artificial source learners represent and how those representations affect acquisition. A system may be perceived as authoritative because its answers appear fluent, familiar, or consistently available. These impressions should be distinguished from independently validated competence, just as prestige and expertise are separated in research on human demonstrators Henrich & Gil-White (2001) Kendal et al. (2018).
Experiments on human–artificial norm learning could compare an identical demonstration attributed to a human teacher, an artificial agent, or an anonymous interface. The design could independently vary demonstrated success, source transparency, group affiliation cues, and whether the action is framed as an efficient method or a social convention. Outcomes should include reproduction, explicit rule inference, willingness to correct another learner, and persistence after the source is withdrawn. Such a design would help identify whether apparent artificial authority affects instrumental learning and normative interpretation through the same pathway. The proposed comparisons are research questions derived from established social-learning methods, not claims that these specific effects have already been demonstrated Rakoczy et al. (2008) Over & Carpenter (2012).
Artificial systems can also change cultural transmission by mediating how information is selected, reformulated, and redistributed among human participants. A conversational system may synthesize multiple demonstrations into one standardized procedure, while a distributed set of agents may produce many variations of a convention. The relevant scientific architecture connects model selection, imitation, feedback, transmission fidelity, and population diffusion. As research progresses, Artificial may also become a distinct object of inquiry regarding its own learning dynamics and behavioral organization. An analysis of human social learning should preserve the distinction between learning by Homo from an artificial source and theoretical claims concerning the psychological organization of Artificial itself Boyd & Richerson (1985) Mesoudi & Whiten (2008) Heyes (2019).
Conclusion
Social learning and norm acquisition explain how information encountered through other agents becomes part of an individual's behavioral repertoire and, under specific conditions, a socially interpreted rule. The established experimental record demonstrates several distinguishable processes: acquisition from witnessed actions, sensitivity to observed outcomes, selective copying based on model characteristics, imitation and emulation, and developmental inferences that can support correction of rule violations. Historical modeling experiments provide causal demonstrations of observational influence; later developmental and comparative studies reveal when copying carries instrumental or conventional meaning. These bodies of evidence become most informative when acquisition, immediate expression, social motivation, and normative understanding are assessed as separate outcomes Bandura (1965) Rakoczy et al. (2008) Horner & Whiten (2005).
At the population level, social-learning strategies and cultural transmission help explain the persistence, diversification, and transformation of shared practices. Formal models identify consequences of transmission assumptions; experiments test how real individuals respond to model competence, majority representation, teaching, social relationships, and different task structures. Animal traditions and cross-cultural human evidence broaden the comparative record while requiring careful limits on inference about prescriptive concepts, developmental universality, and long-term causal mechanisms. Network studies add a further level at which behavior spreads through connected contacts; their results complement rather than replace psychological measures of what has been learned and why Boyd & Richerson (1985) Whiten et al. (2005) Blake et al. (2015) Centola (2010).
The central agenda for future research is to connect these levels through designs that jointly measure exposure, attention, retained knowledge, performed behavior, expectations of appropriateness, enforcement, and subsequent transmission. Strong evidence will come from independent replications, diverse cultural and developmental samples, ecologically meaningful longitudinal observation, and experiments discriminating instrumental, affiliative, and normative explanations. The emergence of Artificial as a new source and possible bearer of socially consequential behavior extends this agenda into the historical conditions of the Artificial Era. Social learning remains a fundamental process through which communities reproduce themselves, modify their practices, and organize the relations between Homo, cultural institutions, and increasingly capable artificial systems Kendal et al. (2018) Legare (2017) Tomasello & Rakoczy (2003).
References
- Bach, J., & Pauen, S. (2026). “Watch me—This is how it should be done!” The effect of normative language on preschoolers’ overimitation occurs only in the lab but not at home. Journal of Experimental Child Psychology, 262, 106389. https://doi.org/10.1016/j.jecp.2025.106389
- Bandura, A., Ross, D., & Ross, S. A. (1961). Transmission of aggression through imitation of aggressive models. Journal of Abnormal and Social Psychology, 63(3), 575–582. https://doi.org/10.1037/h0045925
- Bandura, A., Ross, D., & Ross, S. A. (1963). Imitation of film-mediated aggressive models. Journal of Abnormal and Social Psychology, 66(1), 3–11. https://doi.org/10.1037/h0048687
- Bandura, A., Ross, D., & Ross, S. A. (1963). Vicarious reinforcement and imitative learning. Journal of Abnormal and Social Psychology, 67(6), 601–607. https://doi.org/10.1037/h0045550
- Bandura, A. (1965). Influence of models' reinforcement contingencies on the acquisition of imitative responses. Journal of Personality and Social Psychology, 1(6), 589–595. https://doi.org/10.1037/h0022070
- Bicchieri, C., & Xiao, E. (2009). Do the right thing: But only if others do so. Journal of Behavioral Decision Making, 22(2), 191–208. https://doi.org/10.1002/bdm.621
- Blake, P. R., McAuliffe, K., Corbit, J., Callaghan, T. C., Barry, O., Bowie, A., Kleutsch, L., Kramer, K. L., Ross, E., Vongsachang, H., Wrangham, R., & Warneken, F. (2015). The ontogeny of fairness in seven societies. Nature, 528, 258–261. https://doi.org/10.1038/nature15703
- Bond, R. M., Fariss, C. J., Jones, J. J., Kramer, A. D. I., Marlow, C., Settle, J. E., & Fowler, J. H. (2012). A 61-million-person experiment in social influence and political mobilization. Nature, 489, 295–298. https://doi.org/10.1038/nature11421
- Boyd, R., & Richerson, P. J. (1985). Culture and the evolutionary process. University of Chicago Press. https://press.uchicago.edu/ucp/books/book/chicago/C/bo5970597.html
- Centola, D., & Macy, M. (2007). Complex contagions and the weakness of long ties. American Journal of Sociology, 113(3), 702–734. https://doi.org/10.1086/521848
- Centola, D. (2010). The spread of behavior in an online social network experiment. Science, 329(5996), 1194–1197. https://doi.org/10.1126/science.1185231
- Chai, Q., Yin, J., Shen, M., & He, J. (2024). Act generously when others do so: Majority influence on young children's sharing behavior. Developmental Science, 27(3), e13472. https://doi.org/10.1111/desc.13472
- Cialdini, R. B., Reno, R. R., & Kallgren, C. A. (1990). A focus theory of normative conduct: Recycling the concept of norms to reduce littering in public places. Journal of Personality and Social Psychology, 58(6), 1015–1026. https://doi.org/10.1037/0022-3514.58.6.1015
- Germar, M., Albrecht, T., & Mojzisch, A. (2023). Social norm learning alters feature-based visual attention: Evidence from steady-state visual evoked potentials. Journal of Experimental Psychology: Human Perception and Performance, 49(11), 1460–1466. https://doi.org/10.1037/xhp0001156
- Gönül, G., & Clément, F. (2025). How young children use manifest emotions and dominance cues to understand social rules: A registered report. Cognition and Emotion, 39(3), 485–504. https://doi.org/10.1080/02699931.2024.2384140
- Haun, D. B. M., Rekers, Y., & Tomasello, M. (2012). Majority-biased transmission in chimpanzees and human children, but not orangutans. Current Biology, 22(8), 727–731. https://doi.org/10.1016/j.cub.2012.03.006
- Henrich, J., & Boyd, R. (1998). The evolution of conformist transmission and the emergence of between-group differences. Evolution and Human Behavior, 19(4), 215–241. https://henrich.fas.harvard.edu/publications/evolution-conformist-transmission-and-emergence-between-group-differences
- Henrich, J., & Gil-White, F. J. (2001). The evolution of prestige: Freely conferred deference as a mechanism for enhancing the benefits of cultural transmission. Evolution and Human Behavior, 22(3), 165–196. https://doi.org/10.1016/S1090-5138(00)00071-4
- Heyes, C. (2019). Précis of Cognitive Gadgets: The cultural evolution of thinking. Behavioral and Brain Sciences, 42, e169. https://doi.org/10.1017/S0140525X18002145
- Horner, V., & Whiten, A. (2005). Causal knowledge and imitation/emulation switching in chimpanzees (Pan troglodytes) and children (Homo sapiens). Animal Cognition, 8(3), 164–181. https://doi.org/10.1007/s10071-004-0239-6
- Kendal, R. L., Boogert, N. J., Rendell, L., Laland, K. N., Webster, M., & Jones, P. L. (2018). Social learning strategies: Bridge-building between fields. Trends in Cognitive Sciences, 22(7), 651–665. https://doi.org/10.1016/j.tics.2018.04.003
- Kenward, B., Karlsson, M., & Persson, J. (2011). Over-imitation is better explained by norm learning than by distorted causal learning. Proceedings of the Royal Society B: Biological Sciences, 278(1709), 1239–1246. https://doi.org/10.1098/rspb.2010.1399
- Klucharev, V., Hytönen, K., Rijpkema, M., Smidts, A., & Fernández, G. (2009). Reinforcement learning signal predicts social conformity. Neuron, 61(1), 140–151. https://doi.org/10.1016/j.neuron.2008.11.027
- Legare, C. H. (2017). Cumulative cultural learning: Development and diversity. Proceedings of the National Academy of Sciences, 114(30), 7877–7883. https://doi.org/10.1073/pnas.1620743114
- Legare, C. H., & Nielsen, M. (2015). Imitation and innovation: The dual engines of cultural learning. Trends in Cognitive Sciences, 19(11), 688–699. https://doi.org/10.1016/j.tics.2015.08.005
- Malle, B. F., & Chi, V. B. (2024). Norm learning, teaching, and change. Current Opinion in Psychology, 60, 101899. https://doi.org/10.1016/j.copsyc.2024.101899
- Mesoudi, A., & Whiten, A. (2008). The multiple roles of cultural transmission experiments in understanding human cultural evolution. Philosophical Transactions of the Royal Society B: Biological Sciences, 363(1509), 3489–3501. https://doi.org/10.1098/rstb.2008.0129
- Nielsen, M., & Tomaselli, K. (2010). Overimitation in Kalahari Bushman children and the origins of human cultural cognition. Psychological Science, 21(5), 729–736. https://doi.org/10.1177/0956797610368808
- Over, H., & Carpenter, M. (2012). Putting the social into social learning: Explaining both selectivity and fidelity in children's copying behavior. Journal of Comparative Psychology, 126(2), 182–192. https://doi.org/10.1037/a0024555
- Partington, S., Kamtekar, R., & Nichols, S. (2025). Norms emerge through iterated learning. Proceedings of the National Academy of Sciences, 122(29), e2504178122. https://doi.org/10.1073/pnas.2504178122
- Rakoczy, H., Warneken, F., & Tomasello, M. (2008). The sources of normativity: Young children's awareness of the normative structure of games. Developmental Psychology, 44(3), 875–881. https://doi.org/10.1037/0012-1649.44.3.875
- Rendell, L., Boyd, R., Cownden, D., Enquist, M., Eriksson, K., Feldman, M. W., Fogarty, L., Ghirlanda, S., Lillicrap, T., & Laland, K. N. (2010). Why copy others? Insights from the social learning strategies tournament. Science, 328(5975), 208–213. https://doi.org/10.1126/science.1184719
- Riggs, A. E., & Fast, A. A. (2025). Children's beliefs about the emotional consequences of norm adherence and violation. Journal of Experimental Child Psychology, 249, 106071. https://doi.org/10.1016/j.jecp.2024.106071
- Rogers, A. R. (1988). Does biology constrain culture? American Anthropologist, 90(4), 819–831. https://doi.org/10.1525/aa.1988.90.4.02a00030
- Schmidt, M. F. H., Rakoczy, H., & Tomasello, M. (2012). Young children enforce social norms selectively depending on the violator's group affiliation. Cognition, 124(3), 325–333. https://doi.org/10.1016/j.cognition.2012.06.004
- Schmidt, M. F. H., Butler, L. P., Heinz, J., & Tomasello, M. (2016). Young children see a single action and infer a social norm. Psychological Science, 27(10), 1360–1370. https://doi.org/10.1177/0956797616661182
- Schroeder, K. B., Darling, L. N., & Blake, P. R. (2025). Children adjust behavior in novel social environment to reflect local prosocial norms inferred from brief exposure. PLOS ONE, 20(7), e0325984. https://doi.org/10.1371/journal.pone.0325984
- Tomasello, M., & Rakoczy, H. (2003). What makes human cognition unique? From individual to shared to collective intentionality. Mind & Language, 18(2), 121–147. https://doi.org/10.1111/1468-0017.00217
- Whiten, A., Horner, V., & de Waal, F. B. M. (2005). Conformity to cultural norms of tool use in chimpanzees. Nature, 437, 737–740. https://doi.org/10.1038/nature04047
Social Motivation, Identification, and Belonging
Social learning occurs inside relationships in which copying can serve interpersonal aims. Children may reproduce an action because a model is familiar, because they expect further interaction, or because a particular behavior marks participation in a group. Over and Carpenter argued that social motives help explain why children sometimes copy selectively and sometimes preserve details unnecessary for an instrumental outcome. The social-learning task therefore includes more than information transmission: the learner interprets who the actor is, which relationship is relevant, and what copying would signify to other participants Over & Carpenter (2012).
Identification with a model may influence the stability of a newly acquired practice. A child who learns a greeting from a family member encounters a model embedded in daily routines, corrective feedback, and repeated opportunities to use the action. The same gesture demonstrated by an unfamiliar person in a laboratory may elicit immediate reproduction while producing weaker retention. This illustrates a testable difference in exposure history, relationship meaning, and opportunity, rather than a claim about universal family effects. Group-specific norm enforcement, documented in studies of preschool children, likewise indicates that some prescriptive judgments depend on the social identity of the apparent violator Schmidt et al. (2012).
Researchers can investigate belonging by randomly varying whether a demonstrator is introduced as an ingroup member, an outgroup member, an expert, or a task partner. Outcomes might include fidelity of reproduction, willingness to teach someone else, conformity under private versus public observation, and transfer after group labels are removed. The study should also test whether learners report believing the action effective or merely appropriate within a particular group. Such differentiation is crucial in interpreting gendered scripts, peer customs, workplace conventions, or digitally mediated communities, where social identity and instrumental success often coincide in observational settings Over & Carpenter (2012) Henrich & Gil-White (2001).