Research figure: Albert Bandura · Written by Artificial Sapiens
Albert Bandura's experiments on vicarious reinforcement investigated how the consequences experienced by an observed model change the behavior of the observer. Their decisive contribution was the experimental separation of observational acquisition from the immediate performance of what had been learned. In his 1965 study, preschool children who watched an aggressive film character punished imitated fewer of his actions during an initial test than children who watched the character rewarded or left without consequences. When researchers later offered the children incentives to reproduce the same actions, the differences between groups disappeared. Under these conditions, observed punishment inhibited performance without demonstrating a comparable failure to acquire the modeled responses (Bandura, 1965).
The research is often summarized as learning through another person's reward or punishment. That description identifies the central manipulation, but the experiments address a more precise problem: what an observer learns from a demonstration, what an observer expects to happen after acting, and whether the observer chooses to display the acquired response in a particular test. The work connects the earlier studies conducted by Albert Bandura, Dorothea Ross, and Sheila A. Ross in 1961 and 1963 to the larger social learning program and its later social cognitive development (Bandura, Ross, and Ross, 1961; Bandura, Ross, and Ross, 1963; Bandura, 1977).
What Is Vicarious Reinforcement?
Vicarious reinforcement describes the influence of observing favorable consequences delivered to another person for a particular behavior. An observer may become more willing to perform a similar action when a model receives approval, a material reward, success, or another valued consequence. Its counterpart, vicarious punishment, concerns the inhibitory influence of seeing a model experience an adverse consequence. In the original experimental tradition, both were operationalized by manipulating the observed consequences given to a model and measuring the observer's subsequent behavior (Bandura, Ross, and Ross, 1963; Bandura, 1965).
The expression refers to an effect of another person's consequences on the observer. It should not be mistaken for a reward delivered directly to the observer. Direct reinforcement follows the observer's own behavior, whereas vicarious reinforcement supplies information about a model's action and its outcome. The observer may use that information to anticipate consequences, evaluate the model's behavior, or decide whether acting similarly is worthwhile. Bandura's design put these processes into contact by first manipulating what happened to the model and then separately manipulating incentives offered to the child (Bandura, 1965).
Observed consequences can influence several processes at once. They may attract attention to a model, modify expectations about likely rewards and punishments, signal a social rule, change the model's perceived status, or alter a response's immediate desirability. The original experiment primarily tested the overall behavioral consequence of these manipulations. It did not independently measure every cognitive pathway by which a child interpreted the scene; distinctions among attention, expected outcomes, motivation, and social meaning therefore remain explanatory hypotheses that require their own measurements and experimental contrasts (Bandura, 1965; Bandura, 1986).
Acquisition, retention, and behavioral enactment
Acquisition means that an observer has learned enough of a response pattern to reproduce it under suitable circumstances. Retention involves the maintenance or retrieval of information about that pattern. Enactment, sometimes called performance, means that the observer actually emits the behavior during a given observation period. These processes can diverge. A person can know how to repeat a phrase and choose to remain silent; a child can remember an aggressive act and refrain from displaying it when its consequences appear unfavorable. Conversely, showing a response demonstrates its availability in that setting while leaving open questions about its later persistence or generalization (Bandura, 1965; Bandura, 1977).
This distinction made the incentive test methodologically central. If observing punishment had entirely prevented the acquisition of an action, offering the observer a reward immediately afterward would have had little opportunity to reveal detailed matching responses. If punishment mainly discouraged enactment, an attractive reason to perform could elicit responses that had already been learned. The contrast between the initial free-performance test and the subsequent incentive test is the logical center of the 1965 study (Bandura, 1965).
The Research Program Before 1965
The 1965 study followed a sequence of related experiments rather than constituting the entire Bobo doll research program. In 1961, Bandura, Dorothea Ross, and Sheila A. Ross reported that nursery-school children who observed aggressive adult models subsequently reproduced several modeled actions in a setting without the adult present. The study supported an immediate observational-imitation effect for the tasks and behaviors tested. Its outcomes did not establish the long-term development of aggressive traits or the consequences of repeated exposure outside the laboratory (Bandura, Ross, and Ross, 1961).
The same three collaborators then studied film-mediated aggressive models, comparing ways of presenting behavior through live and symbolic demonstrations. Their 1963 paper on film-mediated aggression addressed whether a model had to be physically present for the observed behavior to influence later action. Another 1963 paper, Vicarious Reinforcement and Imitative Learning, examined the consequences received by the model. Nursery-school children were exposed to an aggressive model who was rewarded or punished, a highly expressive but nonaggressive model, or no model. Children who saw aggression rewarded showed more imitation and preferred the successful aggressive model more than children exposed to punishment. The experiment also reported differences in the effects of inhibitory conditions by child sex and type of nonaggressive example (Bandura, Ross, and Ross, 1963a; Bandura, Ross, and Ross, 1963b).
The coauthorship of the 1963 vicarious-reinforcement study is essential: Albert Bandura, Dorothea Ross, and Sheila A. Ross are all credited on that article. The later 1965 paper, Influence of Models' Reinforcement Contingencies on the Acquisition of Imitative Responses, is attributed to Albert Bandura alone. The distinction matters because the 1963 study demonstrated behavioral differences following a model's reward or punishment, while the 1965 study added a critical test designed to discriminate differences in expression from differences in underlying learning (Bandura, Ross, and Ross, 1963b; Bandura, 1965).
The experimental question left open by 1963
Seeing fewer imitated actions after the model is punished permits at least two explanations. First, observers might pay less attention, encode less, or otherwise acquire fewer of the model's responses. Second, observers might learn the same responses but suppress their expression because the model's punishment makes performing them seem undesirable. A simple postexposure imitation count cannot by itself tell these explanations apart. Bandura's next experiment made the contrast more informative by changing the observer's own incentive after the initial test (Bandura, 1965).
The 1965 Experiment: Question, Participants, and Design
The primary publication appeared in the Journal of Personality and Social Psychology in June 1965, volume 1, issue 6, pages 589–595, under the title Influence of Models' Reinforcement Contingencies on the Acquisition of Imitative Responses. The original paper identifies Albert Bandura of Stanford University as its author. Its hypothesis concerns whether reinforcement delivered to the model changes how much imitation observers perform without equivalently changing how much of the model's behavior they have acquired (Bandura, 1965).
The study involved 66 preschool children, with 33 boys and 33 girls, drawn from the Stanford nursery-school context. Children viewed a filmed adult model performing a prescribed series of aggressive acts directed toward an inflatable Bobo doll. The model's action sequence was shown in conditions that differed in what happened to the model afterward. The use of filmed presentations made the model's visible behavior more comparable across conditions while allowing the consequences to be systematically varied (Bandura, 1965).
Three principal observed-outcome conditions were compared. In the model-rewarded condition, the aggressive performer was praised and rewarded. In the model-punished condition, the performer received an adverse social response and punishment. In the no-consequences condition, the presentation contained the modeled behavior without a comparable reward or punishment. The no-consequences condition was a neutral-outcome comparison for the reinforcement manipulation, not a group that had never watched aggression. This matters when describing what the study controlled: all three conditions involved exposure to an aggressive model (Bandura, 1965).
Observation phase: the same class of modeled actions, different consequences
The children first observed the filmed model's sequence. The experimentally important contrast lay in the consequence segment: reward, punishment, or no visible consequence. The researcher could therefore ask whether different outcomes attached to the displayed behavior influenced children's later matching responses. Because the children themselves had not received those model-contingent consequences, the manipulation concerned learning about another person's outcome rather than directly reinforcing the child's own aggressive act (Bandura, 1965).
The aggressive demonstration also had a specialized scientific function. It contained recognizable, relatively specific actions and expressions that observers could later reproduce. This permitted a performance test focused on matching the model's actions rather than merely judging children as generally aggressive or calm. Such operationalization increases control and observability in the laboratory, while constraining generalization: repeating a scripted act toward a doll under experimental circumstances is narrower than persistent harmful aggression toward other people (Bandura, 1965).
Initial free-performance phase: what would children do without a new incentive?
Following the film, children were placed in a setting in which matching actions could be displayed and recorded. The experiment measured how many of the previously witnessed imitative responses each child emitted before the experimenter introduced a new positive incentive for the child's imitation. This phase registered spontaneous or unprompted enactment in that immediate context. The original report found fewer matching aggressive responses among children who had watched the model punished than among children who saw the model rewarded or saw no consequences (Bandura, 1965).
The contrast is best stated carefully. Observed punishment reduced the measured rate of immediate imitative performance. The initial score alone could not establish that children in the punished-model group had failed to learn the actions. Nor did the report show that rewards necessarily produced greater performance than the no-consequence comparison: the central statistically reported distinction was between punishment and the other two conditions (Bandura, 1965).
Incentive phase: testing for responses that were not initially expressed
After the initial test, all three groups were offered attractive positive incentives contingent on reproducing the behaviors they had watched. In methodological terms, the experiment changed the observer's direct motivational conditions while holding constant the earlier modeled sequence to which each child had been exposed. Children were encouraged to display matching responses they might have acquired even when they had not performed them before the incentive was introduced (Bandura, 1965).
The previously observed between-condition differences disappeared in the incentive test. Under the positive incentive, children who had earlier watched punishment displayed matching actions at a level comparable to children in the reward and no-consequence conditions. This was the decisive empirical result: all three groups demonstrated access to a similar repertoire of modeled actions when motivated to express them (Bandura, 1965).
The word similar remains important. The experiment supports equivalent demonstrable acquisition under its specific procedure and measures; it does not establish that every participant encoded every detail equally, or that all learning processes were identical. A positive incentive is a diagnostic manipulation with limits, not a direct view into memory. Differences in expectations, readiness to act, or other unmeasured variables can influence both phases (Bandura, 1965).
Results and Their Interpretation
The two-stage design generated a pattern that a single observation period could have concealed. In the first stage, punishment of the model was associated with inhibited imitation. In the second stage, incentive offered to the child removed the between-group difference. That sequence supported Bandura's view that a model's reinforcement contingencies exerted a particularly strong influence on performance, while the observed response patterns could be acquired even in the absence of favorable consequences for the model (Bandura, 1965).
The experiment established a localized causal inference because researchers deliberately varied the consequences shown to different groups and then changed the child's incentive during testing. Within the controlled paradigm, the manipulated social outcomes affected measurable behavior. More general causal claims—for example, that watching punishment reliably changes lifelong conduct, that imitation must always occur after observing reward, or that all observational learning is independent of reinforcement—extend beyond the outcome measures of this experiment (Bandura, 1965).
Why the rewarded and no-consequence groups matter
The no-consequence comparison provides an informative reference point. The fact that children in the rewarded and no-consequence conditions both displayed more matching behavior than children who observed punishment indicates that an explicitly favorable model outcome was not necessary for a relatively high level of immediate imitation in this setting. The punished model's outcome was especially influential as an inhibitory signal. Observed reward may have conveyed approval or expected benefit, but the data should not be recast as demonstrating a significant difference between the rewarded and neutral conditions without a reported test supporting that claim (Bandura, 1965).
Why the second test changed the theory
Before the incentive manipulation, low imitation could be misread as low learning. The second test showed why performance is an incomplete index of acquisition: newly available motivation made previously absent matching responses appear. The procedure gave behavioral evidence for latent acquisition without requiring speculative claims about what a child privately believed. Bandura's analysis was influential because it identified an experimentally tractable gap between an observer's capacity to reproduce what was seen and the observer's decision or motivation to reproduce it in a particular situation (Bandura, 1965).
What was actually measured?
The experiment measured overt matching responses to a distinctive filmed behavioral sequence during defined observation phases. It did not measure a child's enduring aggressiveness, moral character, psychiatric status, future antisocial behavior, or willingness to harm a real person in ordinary settings. Aggressive imitation in the task was the operational dependent variable. The study's strength lies in its precision about that variable; its interpretive limits follow from the distance between scripted laboratory play and socially consequential aggression (Bandura, 1965).
Reinforcement, Punishment, and the Meaning of Consequences
In learning psychology, a reinforcer is conventionally discussed in relation to its effect on the future probability of behavior under the relevant conditions. The films in Bandura's experiment displayed socially recognizable outcomes—reward, punishment, or absence of overt consequences—attached to a model's actions. For the observer, those outcomes served as information and social cues that could alter performance. A comprehensive account therefore considers the model's outcome as an observed event and the observer's behavior as a separate measured outcome (Bandura, 1965; Bandura, 1977).
A distinction is also required between vicarious punishment and directly experienced punishment. In the former, the child observes another person suffer an adverse consequence; in the latter, the child personally experiences an outcome contingent on the child's own action. The 1965 test deliberately separated these events. Only the later phase supplied the child with a new direct incentive to reproduce observed responses. This distinction prevents confusion about which individual was reinforced or punished at each stage (Bandura, 1965).
The word punishment in this literature describes a consequence manipulation in a historical experiment. It does not constitute a general recommendation to punish children, display violence, or employ coercive discipline. The research question concerns how observation modifies the expression of behavior. Modern ethical evaluation and evidence-based intervention require separate assessment of outcomes, harms, developmental context, and available alternatives (Bandura, 1965; Malouff et al., 2009).
Theoretical Significance: Learning by Observation
Bandura used the results to challenge the assumption that learning new forms of behavior depends entirely on being personally rewarded for successive response attempts. A child in the 1965 study could later reproduce a complex modeled sequence despite having observed its performer punished. The model provided information about what to do; the consequence helped govern whether the child would do it during the initial test. This distinction formed part of Bandura's social learning account of modeling and guided the movement toward a more elaborated social cognitive theory (Bandura, 1965; Bandura, 1977; Bandura, 1986).
Observation and response representation
The film supplied a perceptible behavioral pattern that could be represented and later reconstructed. Bandura's subsequent account emphasized attention to the model, retention of the observed acts, the observer's behavioral production capabilities, and motivational processes that influence whether those acts are carried out. The 1965 experiment directly bears on the motivational part of that distinction and offers behavioral evidence of acquisition despite differential performance. It should not be treated as an independent experimental validation of every later component or cognitive mechanism of the full theory (Bandura, 1965; Bandura, 1977).
Expectancies and decision-making
Observers may treat a model's consequences as information about what will happen if they themselves act similarly. An adverse result may inhibit enactment because the observer anticipates disapproval, punishment, or other costs. A reward may suggest benefits or social endorsement. Yet the 1965 experiment did not isolate anticipated punishment from other influences of the film, such as affective reactions, identification with the model, or interpretations of an authority figure. These remain reasonable candidate mechanisms, not separately demonstrated mediators of the classic effect (Bandura, 1965; Bandura, 1986).
Relation to later social cognitive theory
In Social Foundations of Thought and Action (1986), Bandura developed a broader social cognitive framework encompassing reciprocal influences among behavior, personal processes, and environmental conditions. The earlier vicarious-reinforcement experiments became important historical evidence about modeling and anticipated consequences within that framework. They predated its mature formulation and should be attributed to the experimental social learning program from which the later theory developed (Bandura, 1986).
Comparing the 1961, 1963, and 1965 Studies
The 1961 experiment by Bandura, Dorothea Ross, and Sheila A. Ross tested whether children reproduced actions observed from aggressive adults when the model was no longer present. Their early-1963 film-mediated experiment examined how live and symbolic presentations affected imitation. Their later-1963 vicarious-reinforcement study manipulated rewards and punishments to models and assessed imitative responses and model preference. Bandura's single-authored 1965 experiment sharpened the acquisition-versus-performance issue with an additional direct-incentive test for the child (Bandura, Ross, and Ross, 1961; Bandura, Ross, and Ross, 1963a; Bandura, Ross, and Ross, 1963b; Bandura, 1965).
These studies form a coherent research sequence, yet their methods and inferential targets differ. Grouping them together under the popular expression Bobo doll experiments is historically understandable, but scholarly interpretation benefits from identifying the author list, specific comparison conditions, dependent measures, and evidentiary contribution of each paper. The 1965 paper is especially associated with the claim that observational learning and observed reinforcement contingencies can affect acquisition and performance in different ways (Bandura, 1965).
The series is distinct from the Little Albert conditioning study associated with John B. Watson and Rosalie Rayner. The latter belongs to an earlier line of classical-conditioning research and involved a different procedure, research question, participant, and authorship. The similarity of the names Albert and Bobo in popular explanations supplies no historical connection between the experiments.
Evidence Beyond the Original Experiments
Bandura's account subsequently motivated studies of modeling with different actions, populations, timescales, and observed outcomes. This literature permits more than a simple declaration that the classic effect was confirmed everywhere. Some studies examine whether watching a punished model suppresses a matching response; others examine how long observational imitation persists, how direct contingencies affect maintenance, or how people use observations to infer rewards. These are related but distinct targets of inference (Malouff et al., 2009; Deguchi, Fujita, and Sato, 1988; Burke et al., 2010).
Meta-analytic evidence on vicarious punishment
Malouff, Thorsteinsson, Schutte, and Rooke (2009) synthesized studies testing whether observers display less of a behavior after seeing a model punished for it. Their meta-analysis included 21 research samples and 876 participants and reported an overall standardized effect, d = 0.58, for reduced behavior after vicarious punishment. The authors described effects across live and filmed models, varying punishment types, different behaviors, and samples of children and adults. This provides broader support for a measurable inhibitory behavioral effect under studied conditions (Malouff et al., 2009).
That pooled outcome concerns the observer's displayed behavior. It should not automatically be interpreted as proving that acquisition remained equal across every study, or that the effect persists indefinitely. Meta-analysis aggregates heterogeneous paradigms and depends on the quality, measurement, and publication characteristics of included investigations. It helps test generality of inhibitory performance effects while leaving the exact causal mediators and long-term behavior changes as separate questions (Malouff et al., 2009).
Evidence on persistence and direct reinforcement
Deguchi, Fujita, and Sato (1988) investigated observational learning and vicarious reinforcement using individual-analysis designs with young children and button-pressing models. Modeling initially produced imitation without direct reinforcement, but that behavior declined relatively quickly. Direct reinforcement of children's own matching responses increased and maintained their imitation; removing it for some participants reduced the behavior, and reinstating it increased imitation again. This study complicates sweeping interpretations by illustrating that acquiring or initially displaying a response is different from maintaining it over repeated opportunities (Deguchi, Fujita, and Sato, 1988).
The 1988 findings are neither a direct reproduction of Bandura's 1965 Bobo-doll design nor a refutation of its principal two-stage result. They examine a different response class and the sustained consequences of direct reinforcement. Together, the studies motivate a layered model of learning, performance, and maintenance in which observational information can rapidly establish a response while ongoing contingencies shape whether it continues (Bandura, 1965; Deguchi, Fujita, and Sato, 1988).
Modern experiments on learning from other people's outcomes
Burke, Tobler, Baddeley, and Schultz (2010) investigated neural mechanisms of observational learning in a separate experimental tradition. Their work studied how participants used another person's choices and outcomes while making decisions, connecting behavioral learning from others with experimentally measured neural signals. It supports the broader empirical relevance of observing outcomes during learning, but its task, measures, and neural interpretation differ substantially from Bandura's preschool imitation experiments. It should not be cited as a direct replication of the 1965 study or as proof of Bandura's entire social cognitive theory (Burke et al., 2010).
Ho, MacGlashan, Littman, and Cushman (2017) analyzed teaching and learning through evaluative social feedback in a computational and theoretical framework. Their work emphasizes how social reward and punishment depend on interpretations of others' goals and intentions. Such accounts provide useful alternative mechanisms through which observed consequences may guide action. They belong to a later modeling tradition and do not rewrite the original historical hypothesis or establish which mediator operated in every child in Bandura's experiment (Ho et al., 2017).
Methodological Strengths and Limits
The 1965 paper's principal strength is its two-stage design. By combining a between-group manipulation of model outcomes with a later motivational intervention for the observer, it addressed an inferential ambiguity that a one-time imitation score would have left unresolved. The matching-action outcome also provided a more specific measure than a diffuse judgment of aggression. The separate no-consequence condition offered a needed reference for interpreting observed reward and punishment (Bandura, 1965).
Sample and setting
The sample consisted of preschoolers recruited in a particular Stanford nursery-school setting, at a time and place with particular social conventions. The experiment studied immediate reactions to short filmed scenarios under laboratory conditions. Generalizing to adolescents, adults, different cultural environments, diverse forms of behavior, and digital media requires independent evidence. The sample and historical context restrict confidence in claims about population prevalence or the magnitude of comparable effects in modern everyday environments (Bandura, 1965).
Construct validity: imitation and aggression
The dependent variable was the child's reproduction of recognizable acts seen in the film. The content was aggressive, but the scientific observation concerned short-term imitation of a modeled script. The task does not establish the origin of all aggression, aggression toward real victims, a stable aggressive disposition, a clinical disorder, or eventual criminal behavior. Similar-looking behavioral outcomes in a laboratory can carry different meanings in unstructured settings. Research on consequential aggression therefore needs additional measures and longer designs (Bandura, Ross, and Ross, 1961; Bandura, 1965).
Measurement and observer expectations
An imitation score depends on which responses are counted as matching and how consistently observers record them. Historical experimental reports should be evaluated for the available coding procedures, condition assignment, statistical contrasts, and the possibility that expectations influence observations. The prominent published result supports a specific condition comparison; it does not by itself supply a modern preregistered analysis plan or establish that all potentially relevant variables were measured. Reanalysis and direct independent replication would permit stronger precision about effect size and reliability (Bandura, 1965).
The incentive-test inference
The incentive test operationalized previously acquired knowledge by inviting children to reproduce the observed acts. The convergence of scores is powerful evidence against interpreting low initial performance as wholesale failure of observational acquisition. Yet the test only reveals behavior that children can and are willing to display in the second condition. Incentives can change attention to test instructions, effort, response thresholds, and willingness to comply. The study therefore supports equivalent revealed repertoires under its testing conditions, while deeper conclusions about memory representations call for separate procedures (Bandura, 1965).
Short-term causation and long-term development
The experiment directly concerns behavior over a short testing interval. Long-term changes in a person's values, habits, or conduct would require prospective designs that repeatedly measure participants after exposure and account for family, peers, context, and baseline differences. Even robust short-term causal effects do not imply inevitable developmental trajectories. This distinction also applies when considering television, streaming media, online creators, or artificial agents as models: each new environment changes the ecological conditions in which observation and consequence information occur.
Mechanism ambiguity
The consequences shown in the film may have operated through anticipated personal outcomes, perceived model status, emotional response, or perceived social norms. Because several were bundled into the same manipulation, the experiment cannot uniquely isolate a single psychological mediator. Modern designs can improve identification by varying the valence of consequences separately from the model's expertise, warmth, visibility, social group, and the observer's direct reward opportunities. This refinement extends the program's logic without claiming that later variables were experimentally measured in 1965.
Applications in Education, Social Behavior, and Digital Environments
The most justified application is methodological and conceptual: what learners visibly do immediately after an example can understate what they retained, and observed consequences can influence their willingness to display an available skill. Teachers, clinicians, coaches, and researchers can use this distinction when designing assessments, demonstrations, feedback, and safe learning environments. An individual may refrain from publicly performing a new behavior yet demonstrate it when the task becomes relevant or the incentives change (Bandura, 1965; Bandura, 1977).
In educational contexts, observing another learner receive feedback can communicate expectations about acceptable or effective responses. Such social information may support participation or inhibit a response, depending on what is modeled and how outcomes are interpreted. Educational claims about improving achievement, skill retention, and transfer require studies that actually measure those outcomes rather than simply importing effect sizes from aggression-imitation research. The distinction between demonstration, motivation, and sustained practice is especially useful for evaluating instruction (Deguchi, Fujita, and Sato, 1988).
In media environments, models appear in entertainment, advertising, social networks, and interactive services. Viewers may notice the likes, criticism, rewards, or sanctions attached to others' actions. Bandura's work provides a historically important conceptual framework for asking whether such observed feedback changes action, even when the observer receives nothing directly. Modern media effects depend on platform design, viewing patterns, population, competing models, and social context; the 1965 laboratory experiment alone cannot quantify causal effects of any particular contemporary platform (Bandura, 1965; Burke et al., 2010).
For Artificial Era research, observed consequences delivered to synthetic characters or artificial agents can be studied as potential social information for human observers. This is an application of established questions about modeling and behavior, not a historical assertion that Bandura experimentally studied current conversational AI systems. Questions about an artificial agent's subjective experience are conceptually separate from measuring a human observer's imitation, expectations, or reward learning. The relevant experiment measures observable human responses to a presented model and its visible outcomes.
Scientific Status and Open Questions
Established within the original experiment: the observed consequence manipulation changed immediate imitative performance, and positive incentives offered to children eliminated the main between-group difference in revealed matching responses. Supported beyond the original study: a meta-analysis of vicarious punishment identified a moderate average inhibitory effect on observable behavior across varied research settings. Limited or contested: the general magnitude of such effects across modern digital environments, their long-term persistence, specific mediating cognitive mechanisms, and direct implications for real-world aggression or clinical intervention (Bandura, 1965; Malouff et al., 2009).
Further research benefits from preregistered multi-site replication, objective and blinded response coding where feasible, diverse participant samples, independent tests of reward versus punishment and no-consequence conditions, and follow-up measures that separate immediate enactment from retained ability and later transfer. Researchers can experimentally vary direct and vicarious incentives while measuring expectations and attention, allowing the theories of why consequences matter to be more sharply distinguished. These proposals follow the inferential structure of Bandura's paradigm rather than being outcomes demonstrated by the original study.
Frequently Asked Questions
What did Albert Bandura find in his 1965 vicarious reinforcement experiment?
Children who saw an aggressive filmed model punished showed fewer matching actions during an initial free-performance test than children who saw the model rewarded or receiving no consequences. When all groups were subsequently offered attractive rewards for reproducing the actions, the between-group performance differences disappeared. The result supported a distinction between acquiring an observed response and choosing to enact it immediately (Bandura, 1965).
Does vicarious reinforcement mean the observer receives a reward?
No. The reward or punishment is first observed being delivered to the model. Direct reinforcement of the observer is a different manipulation. Bandura's 1965 study used both in sequence: children saw consequences for the film model, and later the children themselves were offered positive incentives for matching responses (Bandura, 1965).
Did punishment prevent the children from learning aggressive behavior?
The 1965 results did not support that simple conclusion. Observing punishment reduced immediate performance, but the punished-model group later reproduced the observed actions when offered incentives. The experimental finding concerns the distinction between outward enactment and the response repertoire revealed in a subsequent test (Bandura, 1965).
Was the 1963 vicarious reinforcement experiment written by Bandura alone?
No. Vicarious Reinforcement and Imitative Learning (1963) was coauthored by Albert Bandura, Dorothea Ross, and Sheila A. Ross. The later article on reinforcement contingencies and the acquisition of imitative responses (1965) lists Albert Bandura as its sole author (Bandura, Ross, and Ross, 1963b; Bandura, 1965).
Are the 1965 findings proof that television or social media causes violent crime?
The measured outcome was short-term imitation of filmed behavior in a controlled preschool laboratory task. Determining effects on crime or enduring aggression requires different populations, outcomes, timescales, and evidence. The experiment identifies a mechanism by which observed consequences affect behavior under specified conditions; it does not furnish a direct causal estimate for violent crime (Bandura, 1965).
Is there evidence beyond the original Bobo doll studies?
Yes. A 2009 meta-analysis reported reduced observer behavior after vicarious punishment across 21 research samples. Other experiments studied observational learning and direct reinforcement of imitation. Their findings support aspects of the broader research program while testing different behaviors and outcomes rather than uniformly replicating the 1965 protocol (Malouff et al., 2009; Deguchi, Fujita, and Sato, 1988).
Conclusion
Bandura's vicarious-reinforcement experiments transformed the interpretation of learning from modeled behavior. The 1963 work with Dorothea and Sheila Ross established a foundation for studying how reward and punishment given to a model influence children's subsequent imitation. Bandura's 1965 experiment then asked whether that influence reflected a difference in acquired responses or a difference in their immediate expression. Its two-stage evidence demonstrated how a punished model could inhibit observed performance while children still retained the capacity to produce matching behavior when motivated. That finding remains a foundational methodological insight: experimental measures of action and evidence of acquisition must be interpreted with their motivational circumstances in view (Bandura, Ross, and Ross, 1963b; Bandura, 1965).
Related Articles
Albert Bandura — A comprehensive scientific chapter about Bandura's life, experimental research, theories, and intellectual legacy.
Social-Cognitive Personality Theory: Reciprocal Determinism and Behavior — The broader social cognitive framework connecting personal processes, environmental influences, and behavior.
Learning Theories in Education: Behaviorism, Cognitivism, Constructivism, and Sociocultural Learning — A comparative account of major learning theories and their educational implications.
Albert Bandura's Social Learning Theory (1977 Book) — A close reading of Bandura's influential 1977 monograph.
Albert Bandura's Bobo Doll Experiments — Experimental design, findings, coauthors, and interpretive limits.
References
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
Bandura, A. (1977). Social learning theory. Prentice Hall.
Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Prentice Hall.
Bandura, A., Ross, D., and 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., and Ross, S. A. (1963a). 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., and Ross, S. A. (1963b). Vicarious reinforcement and imitative learning. Journal of Abnormal and Social Psychology, 67(6), 601–607. https://doi.org/10.1037/h0045550
Burke, C. J., Tobler, P. N., Baddeley, M., and Schultz, W. (2010). Neural mechanisms of observational learning. Proceedings of the National Academy of Sciences, 107(32), 14431–14436. https://doi.org/10.1073/pnas.1003111107
Deguchi, H., Fujita, T., and Sato, M. (1988). Reinforcement control of observational learning in young children: A behavioral analysis of modeling. Journal of Experimental Child Psychology, 46(3), 362–371. https://doi.org/10.1016/0022-0965(88)90066-5
Ho, M. K., MacGlashan, J., Littman, M. L., and Cushman, F. (2017). Social is special: A normative framework for teaching with and learning from evaluative feedback. Cognition, 167, 91–106. https://doi.org/10.1016/j.cognition.2017.03.006
Malouff, J. M., Thorsteinsson, E. B., Schutte, N. S., and Rooke, S. E. (2009). Effects of vicarious punishment: A meta-analysis. The Journal of General Psychology, 136(3), 271–285. https://doi.org/10.3200/GENP.136.3.271-286