Testing the developmental foundations of cinematic continuity

To make sense of moving images viewers need to perceive the continuity across film cuts. In the early days of cinema, most films filmed in a single run (no cut) from a static camera. Shortly thereafter filmmakers combined multiple shots to create more compelling visual narratives. A suite of editing conventions that allowed viewers to effortlessly perceive continuity across film cuts emerged through trial and error. Most of these conventions were in common usage by 1918 (Bordwell, Staiger & Thompson, 1985) and they permeate much of visual media -including infant-directed ones- today. One of these conventions is the eye-line match between two juxtaposed shots, which is based on the premise that an audience will want to see what the character on-screen is seeing. A film sequence with an eye-line match begins with a character looking at something off-screen, followed by a cut of another object or person. From a developmental perspective, this refers to gaze following, which typically begins to emerge very early in infancy (D’Entremont et al. 1997; Farroni et al. 2004; Hood et al. 1998; Scaife and Bruner 1975). In an eye tracking study gaze following emerged between 2 and 4 months and stabilized between 6 and 8 months of age (Gredeback et al., 2010). A recent study (McClure, Chentsova-Dutton, Holochwost, Parrott, Barr, 2017) examining gaze following across video chat showed a similar developmental trajectory for gaze-following in video as in the real-world (McClure et. al.,2017). It is not known however if the infants would still be able to follow the gaze of the others on screen if the videos were edited. As a matter of fact, adults who have never previously encountered moving images perceive film shots as individual images (Ildirar & Schwan, 2015). In the present study, we examined the role of film editing on infants' ability of following other's gaze. Twelve-month-old Infants (N=20) and adult controls (N=20) watched videos depicting an actor turning her head toward one of two objects either in a single long shot as in traditional gaze following studies or in two multiple shots edited together, one of which shows the actor turning her head and the other shows the gazed-at object as in commercial infant-directed videos. Participant eye movements were recorded using a Tobii TX300. Each video ended with a still long shot showing the actor and the two objects, one of which had previously been gazed-at. Analysis of gaze behavior during this test shot showed clear gaze following in the adult control group (increased dwell time on the gazed-at object compared to the other object) for both edited and unedited versions. Data collection for the infant sample is on-going but preliminary results indicate that 12 month-olds can successfully followed gaze in the unedited version but are less successful across edits.