The Kuleshov Fallacy

The face has long been regarded as one of the major weapons in the arsenal of cinema—as a tool of characterization, a source of visual fascination, and not least, as a vehicle of emotional expression. Research on emotion from psychology and other disciplines offers a rich resource illuminating the world of expressive behavior on which filmmakers draw, and shape to their own artistic ends, as I discuss in an earlier blog here.

But there is an influential idea in the history of film—part of the lore of film theory, exerting considerable influence among filmmakers—which holds that facial expression is at most of secondary importance in the way that films generate meaning and emotional impact. That idea is the “Kuleshov effect,” named after Lev Kuleshov, one of the heroic generation of Soviet filmmakers who put Soviet cinema at the forefront of the new medium in the 1920s. This vanguard introduced numerous innovations and inaugurated (alongside filmmakers and critics elsewhere in Europe) the tradition of film theory—that is, reflection on the nature of film seeking to identify the unique characteristics and potentials of the emerging art form.

In the early 1920s, Kuleshov conducted a number of informal experiments with his students and colleagues—among them another star-to-be of the Soviet filmmaking scene, V. I. Pudovkin, who would be important in documenting the experiments and disseminating the key conclusions drawn from them.

Image of Lev Kuleshov. Public domain via Wikispace

In one of the experiments, Kuleshov and his team edited an identical shot of the famous actor Ivan Mosjoukine into three different sequences, juxtaposing his face with, respectively, shots of a bowl of soup, a woman in a coffin, and a girl playing with a toy bear. (Descriptions of the experiment offered by Kuleshov and Pudovkin in subsequent years vary in terms of the exact content of the shots – the original footage is lost, though fragments and evidence of the other experiments have survived.) The goal of the experiment was to test—or perhaps more accurately, to demonstrate—the power of editing. Kuleshov found that the emotion viewers attributed to Mosjoukine varied according to the context in which the shot was placed: hunger when juxtaposed with the bowl of soup, sadness with the coffin, and happiness with the child. This affirmed the view among the Soviet film innovators that montage – editing – was the key to the power of cinema. In this way, the experiment made a critical contribution to the movement that would come to be dubbed “Soviet montage.”

But does the Kuleshov effect hold water? Recent attempts to re-test the Kuleshov hypothesis more rigorously suggest that it is a reality (though there is plenty of devil in the detail); those results won’t surprise anyone with expertise as a filmmaker or critic. But without careful handling, Kuleshov’s insight turns into a fallacy. The idea of the Kuleshov effect initiated by the original experiment and that continues to replicate – the Kuleshov meme – is deeply misleading to the extent that it suggests that performative factors, like facial expression, play no significant role in shaping our experience of films. But many filmmakers and critics continue to peddle the idea.

In a generally very insightful exploration of Bernard Herrmann’s films scores, for example, the film composer Howard Goodall says of the sequence in Psycho where Marion Crane drives into the night with the money she’s stolen from her boss: “What’s remarkable about [Herrmann’s cue for the scene] is how much impact it has on the pictures. Without it, what we’re looking at is someone driving along in a car, and there’s nothing dramatic, tense, or worrying about that.” Here the contextual factor invoked by Goodall is the score rather than the editing; but the underlying implication is the same—the face is sufficiently inert or open to interpretation that any role it might play in conveying expressive information is overridden by contextual features, like Herrmann’s score in this case.

Janet Leigh as Marion Crane in Psycho (Alfred Hitchcock, 1960). Images provided by Murray Smith.

But a close look at the sequence exposes the fallacy at work here. Janet Leigh’s performance conveys the emotions of Marion in nuanced but palpable fashion. Her anxiety is evident from a variety of telltale facial expressions, including lip-biting and squinting as night falls and headlights from the oncoming traffic dazzle her. And her facial behavior is of a piece with her bodily demeanor—deep sighs and irregular breathing, restless fidgeting with the steering wheel—conforming to the view of Dacher Keltner that facial expressions need to be understood as part of a multimodal expressive system including the voice and the body. So even if we stripped away Herrmann’s score, Marion’s state of mind would remain tangible from these expressive behaviors.

What is still more striking is a brief passage when Marion’s anxiety wanes and a different emotion takes hold of her: a faint but definite smile is visible on her face as she imagines the moment when her boss discovers her theft (represented by dialogue on the soundtrack).

Janet Leigh as Marion Crane in Psycho (Alfred Hitchcock, 1960). Image provided by Murray Smith.

According to the logic of the Kuleshov effect, this fleeting expressive moment ought to be rendered invisible by the context in which it appears (remember, according to Goodall, without the music all we see is “someone driving along in a car”). But in fact, the force of facial and bodily expression here is such that Leigh’s performance acts in counterpoint to the emotional tenor of Herrmann’s angular, edgy cue. This is only possible because facial and bodily expression are powerful, independent sources of expression in films, working alongside other techniques such as editing and music, not mere handmaidens to these other “medium-specific” parameters.

None of this is to say that Kuleshov didn’t discover, or at least demonstrate and popularize, an important aspect of the art of film. But the power of the Kuleshov effect, and its authority as a discovery deriving from an “experiment,” is often overstated. Part of the problem arises from this very characterization of Kuleshov’s explorations as “experiments.” The idea that artists can engage in experimentation is a common one, and plausible enough in a certain sense. But what’s the relationship between an artistic and a scientific experiment? That is a question that deserves independent consideration.

Featured Image Credit: Janet Leigh as Marion Crane in Psycho (Alfred Hitchcock, 1960). Image provided by Murray Smith.