One can evaluate the behavior in a movie in two different ways.
- According to the internal workings of the film universe. An example would be to praise or criticize behavior according to its trueness to the overall psychological portrait of the character, or to a general perceived social or psychological idea of how people are likely to behave.
- According to the effect of the behavior on the viewer. In this game, it tends to be the filmmaker, not a character, who is pitching, and the viewer, not another character, who is catching.
I've written about this dichotomy before, but it's tricky to fit the two methods of evaluation into a unified field theory. Obviously both approaches can be abused. The "internal" approach, despite the appeal to the authority of the soft sciences, is no more or less likely to get bogged down in subjectivity than the more obviously subjective "external" approach. You'd think that it would mean something when a critic says "As a former construction worker, I can testify that the film's portrait of construction workers is accurate." But one learns in one's youth that such statements are 100% subjective and have no bearing on anything at all.
I'm tempted to talk about these methods in terms of "two-ness," which, in this case, would mean meeting the criterion of internal plausibility while at the same time creating a worthwhile external effect. And this is certainly the ideal of a kind of classical storytelling that adheres to notions of social or psychological realism. At the least, verisimilitude and observational insight will always be a valuable arrow in the cinema's quiver.
However, it doesn't do to beat a film with the stick of verisimilitude. When we like a bit of abstract behavior, we forgive its departure from documentary realism because the abstraction gives us something valuable. It's only when we don't get anything from an abstraction, or when we get something we don't like, that we are tempted to say "No one behaves like that." And so the internal approach isn't a completely independent criterion: to make a just criticism of a failure of internal coherence, we have to take external factors into consideration. "No one behaves like that" is never a valid condemnation when taken in isolation from other factors.
More and more I feel that the external approach to evaluation is the larger and more philosophical viewpoint, the one that provides context for issues of internal verisimilitude. And yet it's rare to hear people talk about a movie as if it's a moment-by-moment feed of information and pleasure from the filmmaker to the audience: we are much more likely to try to praise or condemn a movie according to whether we believe the characters, even though we know that the characters are merely the filmmaker's tools.
The main pitfall of external evaluation is obvious: it's not immediately obvious why one viewer's response to a filmmaker's stimuli should be of any use to a different viewer. Any attempt to identify objective elements of form that create one's subjective responses is highly likely to devolve into rationalization, even when one brings some rigor to the process of identification. But hey - if it were easy to talk about art, everyone would do it. It's by no means impossible to argue that the delivery of certain experiences is inscribed into a film's form, even if many others don't receive the experience, even if one doesn't receive the experience oneself. There's something up there on screen that is the same for all of us, and that's where the job begins, even if one must proceed with the utmost caution in trying to build a model of pleasure delivery upon the relatively solid foundation of formal analysis.