The question
By the late 1950s, many psychologists explained learning through reward and punishment: a behaviour is learned because it pays off. Albert Bandura thought this left out something obvious from everyday life. Children pick up whole routines of words and gestures simply by watching other people, often without ever being rewarded for them.
An earlier study by Bandura and Huston had shown that children readily imitate an adult who is still in the room. The sharper test, Bandura and his colleagues argued, was whether children would reproduce what they had seen in a new place, after the adult had left. And because they chose aggression as the behaviour, the study also spoke to a lively debate of the time: does watching aggression drain it away, or teach it?
How it was done
The 1961 study. Seventy-two children from the Stanford University Nursery School took part, 36 boys and 36 girls, aged between 37 and 69 months. Before the study, the experimenter and a nursery teacher rated each child on how aggressive they usually were, so that the groups could be matched.
Each child was brought into a room and given a quiet activity: making pictures with potato prints and stickers. In another corner, an adult played with construction toys next to a mallet and an inflatable Bobo doll about 1.5 metres (5 feet) tall, weighted so that it pops back up when knocked over.
- Aggressive model. After about a minute, the adult turned on the doll with a set of unusual, easy-to-recognise acts: sitting on it and punching its nose, hitting it on the head with the mallet, throwing it in the air and kicking it around the room, each repeated about three times, with phrases such as “Pow”.
- Nonaggressive model. The adult kept playing quietly and ignored the doll.
- Control group. Twenty-four children saw no model at all.
Half of the children in each model condition watched a man, half a woman, and half saw a model of their own sex. The session lasted about 10 minutes.
Next came a deliberate provocation, used for every child. In a second room with attractive toys, the experimenter let the child start playing, then said these toys were being kept for other children. The child was then taken to a third room containing a smaller Bobo doll, a mallet, dart guns and a tetherball, along with crayons, dolls, a tea set and toy animals. For 20 minutes, observers behind a one-way mirror recorded what the child did in every five-second interval.
The 1963 study. The 1963 report compared four groups of 24 children from the Stanford nursery (96 in all). Two groups, the live model and no model, were the children already tested in 1961. One group saw a live adult attack the doll, as before. A second saw a colour film of the same adults. A third saw a film on a television set of a woman dressed as a black cat, “Herman the Cat”, attacking the doll on a cartoon-like set with music. The fourth group saw nothing. All children went through the same frustration step and 20-minute test.
The 1965 study. Bandura alone then asked what happens when children see what the model gets for the behaviour. Children watched a short film of a man attacking the doll and then being rewarded with praise and treats, being scolded and punished, or face no consequences. Afterwards, the children were first observed playing freely. Then they were offered small rewards, such as stickers and juice, for every act of the model they could show.
What came out
1961: watching was enough. Children who had watched the aggressive adult reproduced many of the model’s specific acts and phrases. Children in the other two groups almost never did; around 70% of them scored zero on these measures. About a third of the children who had seen the aggressive model even repeated the model’s harmless remarks. None of the other children did.
Other findings were more mixed:
- Boys copied more physical aggression than girls. Verbal copying did not differ.
- The male model had the stronger effect, especially on boys. Children’s comments suggested why: they saw a man hitting the doll as normal, and a woman doing the same as out of place.
- Watching the calm adult had its own effect. Those children spent more than twice as long sitting quietly as children who had seen aggression, and those who saw the calm man were less aggressive than controls on several measures.
- Not everything was copied. Punching the doll and playing with the guns did not differ between groups.
1963: screens worked too. Average total aggression was 83 after the live model, 92 after the film and 99 after the cartoon, compared with 54 in the group that saw no model. All three model groups were significantly more aggressive than controls, and they did not differ significantly from each other. The live model produced significantly more exact copying than the cartoon. Some imitation appeared in about 88% of children who saw a live or filmed adult, and in 79% of those who saw the cartoon. The authors concluded that watching filmed aggression “substantially increases rather than decreases the probability of aggressive reactions”, the opposite of what the catharsis idea predicted.
1965: learning is not the same as doing. Children who saw the model punished copied fewer of its acts in free play, an effect the paper attributes mostly to the girls. Once children were offered rewards for showing what the model had done, the gap between the groups largely closed. Bandura’s conclusion was that the children had learned the behaviour just by watching; what they saw happen to the model mainly decided whether they chose to perform it.
What really happened (replications and critiques)
The core finding holds up. The basic result, that children quickly reproduce new acts they have just watched, appears across all three studies and in later work. In a small 1971 master’s study at the University of Nebraska at Omaha, with 36 kindergartners, offering an incentive likewise erased group differences in how much the children imitated a televised model. In 1973 B. H. Kniveton reported that the influence of a filmed model could still be seen four to five months later.
But is hitting Bobo aggression? This is the most common objection. Bobo is built to be knocked down and bounce back, so children may have taken hitting it as the obvious way to play rather than as aggression. Hitting a toy is a long way from hurting another child, and the studies did not measure that. In 2022 the Athens-based researchers Evangelia Galanaki and Konstantinos Malafantis went further, arguing that much of what the children did may have been play, and a response to a strange, arousing situation. They acknowledge that their psychoanalytic reading is not supported by research data.
Novelty and expectations. Popular summaries cite a claim, attributed to Cumberbatch (1990), that children who had never seen a Bobo doll were far more likely to copy the model. We could not locate the original source. Another concern is that the setup may simply have shown children what adults expected of them.
An artificial setting. The media scholar David Gauntlett has argued that laboratory measures of this kind, such as aggression toward an inanimate object like the Bobo doll, cannot stand in for how people actually treat each other. He also points to research suggesting that the presence, appearance and gender of an observer can change how children behave.
Who scored the children? The male model scored all the sessions. According to the paper, he did not know which group a child was in, except for the sessions in which he himself had been the model, though the authors admit that children who had seen the aggressive model were easy to recognise from their behaviour. A second observer independently scored half the sessions, and the two agreed closely. The sample was narrow: children from a single university nursery school, tested immediately afterwards.
Ethics
The 1961 report does not mention parental consent, a debrief, or any check on the children afterwards. Every child, including those in the control group, was deliberately frustrated by being told the toys they had just started playing with were meant for others. The experimenter stayed in the test room partly because some children would not remain there alone or wanted to leave early.
Today it would need documented parental consent, the children’s own agreement, a clear way to stop and an explanation at the end. Whether a review board would still approve exposing three-year-olds to an adult attacking a doll, and then frustrating them on purpose, is genuinely open to debate.
What it means today
The Bobo studies changed how psychology thinks about learning. They showed that children can acquire new behaviour simply by watching, with no reward at all, and that what they see happen to others shapes what they choose to do. Bandura built on this to develop his social learning theory.
They are still cited in debates about television and video games, but cannot settle them on their own: they measured short-term play with a toy, not harm to people.
Some reasonable takeaways go beyond what Bandura tested:
- Children notice and store far more than they repeat on the spot, so “they didn’t copy it” doesn’t mean “they didn’t learn it”.
- How adults react when someone behaves badly, on screen or in the room, may matter as much as the behaviour itself.
- Calm models teach too: in 1961, children who watched a quiet adult were themselves quieter.
Would you have passed?
Coming soon. Predict the results first: which 1963 group was most aggressive, and did seeing the model punished stop children from learning? Then compare your guesses with the findings. No aggressive footage is shown.
Questions people ask
What happened in the Bobo doll experiment?
In the 1961 study, nursery-school children watched an adult either attack a large inflatable Bobo doll or play quietly. After being mildly frustrated, each child was left to play in another room with a smaller Bobo doll. Children who had seen the aggressive adult copied many of the adult's specific acts and words; the others almost never did.
Did it matter whether the model was live, on film or a cartoon?
Not much for overall aggression. In the 1963 study, children who saw a live adult, the same adult on film, or a person dressed as a cartoon cat were all more aggressive than children who saw no model, and the three groups did not differ significantly. The live model produced more exact copying than the cartoon.
What did the 1965 study add?
It showed the difference between learning and doing. Children who saw the model punished copied less. When they were then offered rewards for showing what the model had done, the gap between the groups all but vanished: they had learned the behaviour, but held back from performing it.
Does the Bobo doll experiment prove that violent media make children violent?
No. It shows that young children quickly pick up and reproduce aggressive acts they have just watched, in a playroom, with a toy built to be knocked over. Whether and how much this carries over to real aggression toward people, over months and years, is a separate and still debated question.
Sources
- 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. doi: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. doi:10.1037/h0048687
- Bandura, A. (1965). Influence of models' reinforcement contingencies on the acquisition of imitative responses. Journal of Personality and Social Psychology, 1(6), 589–595. doi:10.1037/h0022070
- Adams, G. R. (1971). A test of the 'contiguity' and 'generalized imitation' theories of social modeling processes (Master's thesis). University of Nebraska at Omaha. core.ac.uk/download/232754995.pdf
- Kniveton, B. H. (1973). The effect of rehearsal delay on long-term imitation of filmed aggression. British Journal of Psychology, 64(2), 259–265. doi:10.1111/j.2044-8295.1973.tb01350.x
- Gauntlett, D. (2006). Ten things wrong with the media 'effects' model. In C. K. Weaver & C. Carter (Eds.), Critical readings: Violence and the media. (Original version 1998, in R. Dickinson, R. Harindranath & O. Linné (Eds.), Approaches to audiences. Arnold.) www.davidgauntlett.com/wp-content/uploads/2018/04/Ten-Things-Wrong-2006-version.pdf
- Galanaki, E., & Malafantis, K. D. (2022). Albert Bandura's experiments on aggression modeling in children: A psychoanalytic critique. Frontiers in Psychology, 13, 988877. doi:10.3389/fpsyg.2022.988877
- Dillon, R. F., & Dillon, A. D. (2024). Bobo doll experiment. EBSCO Research Starters. www.ebsco.com/research-starters/health-and-medicine/bobo-doll-experiment
- McLeod, S. Bandura's Bobo doll experiment on social learning. Simply Psychology (page updated 2026). www.simplypsychology.org/bobo-doll.html