12-2: The Bobo Doll Studies & Media Violence
Psychology of Learning
Module 12: Observational Learning 2
Part 2: The Bobo Doll Studies & Media Violence
Looking Back
Part 1 examined biological foundations of observational learning—evolutionary advantages, mirror neurons providing neural substrate for action understanding, & implications for autism. Now we turn to one of psychology’s most famous experimental series demonstrating observational learning of aggression.
Aggressive Behavior: A Concerning Application
A father loved to watch wrestling on TV with his young son. The boy began imitating exactly what he saw—bodyslams on stuffed animals, elbow drops on pillows, flying kicks at furniture. His parents initially found this amusing. But the imitation escalated. The boy began applying wrestling moves to his younger sister. He attempted to place his baby brother in a headlock. What began as seemingly innocent imitation became genuinely dangerous behavior. This scenario motivated Albert Bandura’s groundbreaking research on observational learning of aggression.
Transmission of Aggression Through Imitation of Aggressive Models
APA Citation: Bandura, A., Ross, D., & Ross, S. A. (1961). Transmission of aggression through imitation of aggressive models. Journal of Abnormal & Social Psychology, 63(3), 575–582.
This landmark study fundamentally changed understanding of how children acquire aggressive behaviors. Before Bandura’s research, dominant theories emphasized that aggression resulted from frustration or was reinforced through success. Bandura proposed a radically different mechanism: children could learn complex aggressive behaviors simply by observing models, without direct personal frustration or reinforcement.
Variables & Experimental Design
The experiment manipulated three independent variables in a factorial design. Type of model had three levels: aggressive model who attacked the Bobo doll, nonaggressive model who ignored it, or no model control condition. Gender of model was either male or female adult. Gender of child was either male or female participant. The dependent variable was aggression, operationalized through multiple behavioral measures: imitation of physical aggression, imitation of verbal aggression, partially imitative behavior, & nonimitative aggression.
Children in the modeling conditions were matched for prior aggressiveness, rated on four five-point scales for physical aggression, verbal aggression, aggression towards physical objects, & ability to inhibit aggression. Multiple raters achieved interrater reliability of r = .89.
Procedure
Each child was brought into an experimental room. The child sat in a corner with potato prints & stickers while a model was taken to the opposite corner containing a Tinker-toy set, mallet, & 5-foot inflated Bobo doll. The nonaggressive model played quietly with Tinker-toys & ignored the Bobo doll. The aggressive model played with Tinker-toys for 1 minute, then spent the remaining time aggressing toward the Bobo doll—sitting on it, punching it, hitting it with a mallet, tossing it in the air, & kicking it while making verbal statements: “Hit him down,” “Throw him in the air,” “Kick him,” “Pow!”
To create frustration, children were moved to a room with very attractive toys, allowed to play for 2 minutes, then told: “These are my very best toys, not just anyone can play with them. I need to save them for other children.” During the test phase, children were brought to a room containing aggressive toys (Bobo doll, mallet, dart gun) & nonaggressive toys (tea set, crayons). Children were left for 20 minutes while raters blind to condition observed through a one-way mirror.
Results
Children who observed aggressive models showed dramatically more imitative aggression than children in control conditions—roughly 20 times more imitative physical aggression than the nonaggressive model condition. Boys displayed much more physically aggressive behavior than girls. Children were more likely to imitate same-gender models. Boys showed particularly high imitation of male models for physical aggression. When the aggressive model appeared on video rather than being actually present, children displayed less imitative aggression—suggesting live models are more powerful than filmed models.
Model-Rewarded Versus Model-Punished Conditions
In subsequent studies (Bandura, 1965), adult models were either rewarded or punished after their aggressive behavior. In the model-rewarded condition, a second adult praised the model: “You’re a strong champion! Your superb aggressive performance deserves a generous treat!” & gave candies & soft drinks. In the model-punished condition, the second adult reprimanded the model: “Hey there, you big bully! You quit picking on that clown!” & spanked the model with a rolled-up magazine.
Children who saw the rewarded model displayed more spontaneous imitation than children who saw the punished model. Vicarious reinforcement increased imitation; vicarious punishment decreased it. Observers formed expectancies that they would receive similar consequences for similar behaviors.
The Learning-Performance Distinction
Crucially, when experimenters later offered children rewards for reproducing the model’s behaviors (“I’ll give you candies if you can show me what the model did”), children in BOTH conditions demonstrated equivalent knowledge of the aggressive behaviors. They had learned equally well through observation regardless of whether the model was rewarded or punished. Children who observed punished models could reproduce aggressive acts just as accurately when incentivized.
This demonstrated Bandura’s key learning-performance distinction—observational learning occurs regardless of model consequences through attention, retention, & representation formation, but motivation to perform depends on expected outcomes. Vicarious reinforcement increases spontaneous performance; vicarious punishment decreases spontaneous performance. But the behaviors are learned either way, stored in memory as potential responses.
Implications for Media Violence
This finding has profound implications for media violence. Children observing violent media may be learning aggressive behaviors even when protagonists are punished. The common assumption that showing violence being punished prevents imitation is only partially correct—punishment reduces immediate performance but doesn’t prevent learning. Media violence exposure may create a reservoir of learned aggressive behaviors available for activation under conducive circumstances.
Violence in the Media
Cultural spillover theory proposes that levels of acceptance of violence in media are reflected by levels of acceptance of violence in society—violence observed in media “spills over” into increased societal violence through observational learning & normalization of aggressive behavior (Baron & Straus, 1989). The theory operates through two mechanisms: direct observational learning of aggressive behaviors & gradual normalization reducing inhibitions against violence. Each exposure slightly shifts perceptions of what constitutes normal behavior. Violence becomes familiar, routine, unremarkable.
Research on media violence has produced thousands of studies over six decades. Correlational studies find that children who watch more violent television tend to be more aggressive. Longitudinal studies by Eron & Huesmann (1986) found that boys who watched the most violent television at age 8 were more likely to get in trouble with the law as teenagers, more likely to be convicted of serious crimes as adults, more likely to use violence to discipline their children, & more likely to treat their spouses aggressively.
However, a recent comprehensive meta-analysis by Kim (2024) examined the replicability of media violence effects & found that while effects exist, they are smaller than earlier meta-analyses suggested. Effect sizes were modest, & publication bias may have inflated earlier estimates. The relationship between media violence & aggression is real but more nuanced than dramatic headlines suggest. A longitudinal study in China (Guo et al., 2024) confirmed that media violence exposure predicts later aggression, but effects were moderated by parental involvement & media literacy education.
Video Game Violence
Video games add another dimension to violence, as the person playing the game actually performs virtual acts of violence. The tragic school shooting at Columbine High School was blamed by some on the video game Doom. Anderson, Gentile, & Buckley (2007) reported increased aggressive behavior following exposure to violent video games, even for short periods, in children previously rated as either aggressive or nonaggressive.
Active participation in video games distinguishes them from passive television viewing—players perform violent actions themselves, potentially strengthening learning through behavioral rehearsal. Desensitization may also occur, with reduced emotional responsiveness to violence following repeated exposure. Normalization gradually shifts perceptions of what constitutes acceptable behavior.
A systematic review by Rojas-Jara & colleagues (2023) examined screen violence effects on children’s & adolescents’ mental health, finding consistent associations between violent media exposure & aggressive cognitions, though effects on actual aggressive behavior were more variable. The review emphasized the importance of parental mediation & media literacy interventions.
Methodological Considerations
Interpretation of media violence research requires careful consideration of methodological limitations. Selection effects suggest that correlations between media violence & aggression may reflect aggressive individuals preferring violent media, rather than violent media causing aggression. Reciprocal causation is likely—media violence exposure may increase aggression, while aggressive individuals may seek out more violent content. These bidirectional relationships make causal claims difficult.
Nevertheless, experimental studies (like Bandura’s) demonstrate that exposure to modeled aggression can cause increased aggression in laboratory settings. The practical question is how these laboratory effects translate to real-world outcomes across diverse populations & contexts.
Prosocial Media Effects
On a more positive note, correlations also exist between prosocial television programs & prosocial behavior in children. Programs like Sesame Street & Mister Rogers’ Neighborhood that model positive values demonstrate that observational learning can promote beneficial behaviors. The same mechanisms that enable learning of aggression enable learning of kindness, cooperation, & empathy. Unfortunately, violent content significantly outweighs prosocial content in most media environments.
Rating Systems & Parental Responsibility
The movie, television, music, & video game industries have introduced rating systems to give consumers informed choice. Beginning in 2000, television manufacturers were required to install V-chips allowing consumers to control what children watch. However, critics argue that rating systems serve primarily to shift blame from content producers to consumers, particularly parents. The ratings exist but are often ignored; the V-chip is available but rarely used. Effective protection of children from violent media requires active parental involvement, not just technological solutions.
Looking Forward
Part 3 explores broader applications of observational learning including achievement motivation, self-regulation, & cultural transmission. We examine how children learn standards for self-reinforcement, develop delay of gratification, & transmit complex cultural traditions across generations through social learning.