Does the Kuleshov Effect Really Exist? Revisiting a Classic Film Experiment on Facial Expressions and Emotional Contexts
Anna Rédei Cabak, Daniel Barratt & Joost van de Weijer (CCS, Lund)
In the early days of cinema, the Soviet filmmaker Lev Kuleshov conducted an experiment that has become part of the mythology of film history. Legend has it that Kuleshov combined a close-up of the Russian actor Mozhukin’s neutral face with a variety of different emotional contexts, including a child playing with a doll, a dead woman in a coffin, and a bowl of soup. The viewers of the three film sequences were reported to have perceived Mozhukin’s face as expressing happiness, sadness, and hunger (or thoughtfulness) respectively. It is not clear, however, whether or not the so-called ‘Kuleshov effect’ really exists. Kuleshov’s original film footage is lost and two recent attempts at replication have produced either conflicting or unreliable results.
The current paper describes an attempt to replicate Kuleshov’s original experiment using an improved and extended stimulus set in conjunction with an improved experimental design. In a behavioural and eye tracking study conducted by the authors, 36 participants were each presented with 24 film sequences. Each film sequence comprised an image of a person’s neutral face, followed by an image of an object/event, followed by another image of the person’s neutral face. In line with Kuleshov’s original experiment, we included a happiness condition, a sadness condition, and a hunger condition with equivalent stimuli. To broaden the scope of the experiment, and to increase the overall number of trials, we added a further three conditions: namely, a fear condition, a desire condition, and a null (no-context) condition. For each film sequence, the participant was asked: to rate the valence of the depicted person’s emotion; to rate how aroused the person appeared to be; and to identify the type of emotion that the person was feeling. The participant’s eye movements were recorded throughout the experiment using a remote eye tracker. The results suggest that some sort of Kuleshov effect does in fact exist. For the different emotional conditions, the participants tended to choose the appropriate emotion more frequently than the alternative options. The answers to the valence and arousal questions also went in the expected directions.