Some behaviorists such as N.E. Miller and J. Dollard claimed that imitation is really just a special form of operant conditioning. Operant conditioning describes a situation in which reinforcement (as punishment or reward) makes a behavior more likely to happen. According to Miller and Dollard, when a child imitated his mother, he was only responding to her behavior as a cue and then reinforced as with any other behavior. In fact, their "follow the leader" experiment demonstrated that imitation and non-imitation could both be increased through reinforcement.
It is important to understand why other psychologists such as Albert Bandura criticized Miller's theory of social learning. First of all, imitated behaviors do not always immediately follow the example behavior. Secondly, Miller and Dollard did not describe what caused the very first imitation. Some behaviorists tried to solve this question by suggesting that the first imitation was a generalization--in other words, a behavior only slightly different from one that had been reinforced in the past. For instance, if a teacher reinforced a student for helping a younger child tie his shoes, that same student might also imitate her father helping a sibling put on a coat.
However, by the 1960s, Albert Bandura believed that this idea of generalization remained insufficient and decided to conduct the "Bobo Doll Experiment." In this experiment, three groups of children watched an adult act aggressively toward a blown-up doll. In the case of the first group, the adult was rewarded for his behavior. For the second group, he was punished. And, for the third group, the adult received no consequences. The children were then themselves given the chance to interact with the doll. Bandura's results showed that children who watched the rewarded model showed much greater levels of aggression than those who saw him punished. However, when the children were later reinforced themselves for aggressive behavior, they all began to imitate the words and punches of the adults. Therefore, Bandura concluded that learning of new behavior required only observation, but that display of the behavior depended on either the expectation of reinforcement or reinforcement itself. He stated that attention to the model, memory of the behavior, the physical ability to perform the behavior, and motivation all play roles in whether a person will imitate the behavior.
Walter Mischel conducted studies in the 1970s that further demonstrated that various circumstances affect imitation differently. First of all, models that have given out a lot of reinforcement in the past and dominant models are more likely to cause imitation. Secondly, people are more likely to imitate closer friends and family with whom they have a more secure attachment. They are also more likely to imitate people similar to them in age, gender or interests.