Learning from Models

The doctrine this theme ran up against was the strictest of its day. Behaviourism explained behaviour from consequences: what is reinforced occurs more often, and whoever does nothing learns nothing either. Watson had conditioned fear, Skinner had surveyed the schedules of reinforcement — a closed building with no room in the plans for mere watching.

Albert Bandura’s objection was an everyday observation with explosive force: children acquire language, manners and bad habits at a pace no reinforcement schedule can supply, and they display actions they have never performed themselves and for which they have never seen a reward. The Bobo doll turned this into an experiment: whoever had watched an adult strike the doll struck it too — in the same movements, with the same phrases.

The decisive move came in 1965 and refuted behaviourism in its own language. Children who saw the model punished for striking imitated it less often — offer them a small reward and suddenly all could do it equally well. Acquisition and performance are two different things: all had learned; only those who saw a reason had shown it. The whole programme stands on this distinction.

The neighbourhood of the two folders is no either-or. Skinner’s shaping builds behaviour in steps of approximation; Bandura’s modelling skips the steps where an example suffices — and behaviour therapy uses both to this day: whoever treats a phobia performs the feared act first, lets the patient watch and then join in. In controlled comparisons this participant modelling beat pure confrontation.

Reception loaded more onto the experiment than it can carry. In the quarrel over media violence the doll became a weapon; the meta-analyses find small, scattered associations, and the registered studies on video games have shrunk the early effects further. Bandura himself held both abbreviations to be wrong: whoever derives a ban from his findings overrates them; whoever declares them refuted confuses small effects with none.

Out of the same programme grew the most-measured concept in modern psychology: self-efficacy, the conviction of being able to manage a particular task under particular circumstances — with one’s own success as the strongest source and the observed model as the second. And the largest application of model learning runs not in the laboratory but on television: serial characters who demonstrate what an audience is meant to learn, consequences included. The sentence that carries the theme is narrower than the headline and more durable: we also learn what we have only watched. Whether we do it is decided by something else.