Cognitive Daily

Literary theory is being influenced more and more by research in cognitive psychology, and as the previous article I discussed showed, psychology research is also influenced by theory. Today’s article, “Generating Predictive Inferences While Viewing a Movie” (Joseph P. Magliano, Northern Illinois University, and Katinka Dijkstra and Rolf A. Zwaan, Florida State University, in Discourse Processes, 1996), is another example.

In much film theory, the key techniques used to tell a story are mise en scène, montage, cinematography, and sound. Mise en scène is everything used to create the set, including costuming and stage directions. Montage is simply the techniques used to edit film shots together. Cinematography is the technical craft of shooting a film: the camera, lens, film, filters, framing, etc. Sound, of course, includes music, sound effects, and dialog. Film theorists like David Bordwell have suggested that a combination of several of these devices is more likely to have a significant impact on film viewers than one device employed on its own.

Magliano and his colleagues wanted to test Bordwell’s theory, so they devised two experiments. In the first one, college students watched the over-the-top James Bond flick Moonraker on a VCR, and were encouraged to stop the tape and write down predictions as to what would happen next, whenever such a thought occurred to them. For example, when “Jaws” fell out of a plane without a parachute, and then the film cut to a circus big top, many viewers “predicted” that he would fall into the tent. Magliano et al. eliminated predictions made by only one viewer, and still counted 98 different predictions. They then analyzed the predictions to find the most common ones and see what preceded them in the movie. They redefined Bordwell’s categories into 5 “types of support” for predictions: Mise en scène, Montage, Framing, Music, and Dialog. 57 of the 98 were supported by more than one category (e.g. Montage and Framing). Montage was the most common support for a prediction, with all the other categories accounting for roughly the same number of predictions.

Then the looked at each individual prediction and analyzed what categories were most likely to generate a prediction. Perhaps most surprisingly, Music was unrelated to prediction generation. For all the hype about the importance of music in film, it seems unrelated to whether a viewer will make a prediction about what will come next. What did matter, perhaps less surprisingly, was the number of cinematic devices supporting a given prediction. The more devices, the more likely a viewer was to arrive at that prediction.

In their second experiment, Magliano et al. wanted to know if viewers would make the same predictions without being asked specifically to make predictions. They took another group of college students and showed them the film again. This time, instead of asking them to make predictions, they simply stopped the film at moments when predictions were made by the first group, and asked the participants to write down whatever came to mind. They also stopped the film at several random points where no predictions were made before.

They found that viewers were indeed still making predictions, and they were more likely to make the predictions that the first group had made when the prediction was supported by more than one cinematic device. The effect was dramatic:

Percent of viewers making the same prediction as in Experiment 1
Multiple sources of support
One source of support
No source of support

Magliano et al. realize that these devices are used intentionally by the filmmakers—so therefore the filmmakers must want viewers to make predictions. Why? They suggest that making predictions helps keep viewers involved in following the film. Taking the example of Jaws falling into the big top, filmmakers were so certain that viewers would make the prediction that they didn’t even show Jaws actually landing. They showed him falling without a parachute, then began playing circus music, then cut to the big top, and back to Jaws, and finally back to the inside of the tent, but then moved on to another scene. But viewers were not surprised to see Jaws show up again later in the film.

In Bond films, I think there might be another reason the filmmakers want viewers to make predictions: they want viewers to feel sophisticated. Part of the appeal of the Bond film is that viewers are supposed to be excited by the lifestyles of the wealthy socialites depicted in the film. By encouraging them to make predictions, the filmmakers invite viewers to be a part of that lifestyle; to live it vicariously through Bond.

While some may believe this research to be somewhat anticlimactic since it only supports Bordwell’s theory of film narrative, I should point out that in Monika Suckfull’s work that I discussed previously, the film theorist’s view was not supported. “Unsurprising” results in psychological research often tell us as much as surprising ones do.