Faces in shadows: silhouette light, underlight and toplight elicit increased early posterior negativity.

One key aspect of film lighting, and light in general, is its direction and how it illuminates people and other objects of attention. This research article presents the results of a pilot EEG experiment that studied the emotional responses of nine test subjects to photographs of an expressionless human face lit from varying directions. The aim of the study was to examine, how the direction of the main light source illuminating the face-the so-called 'key light' in filmmaking-would affect the test subjects' subliminal-level emotional response before any conscious emotional processing takes place. EEG studies on how facial lighting affects the viewers' subliminal emotions have not been reported in academic literature but, on the other hand, facial expressions and other emotion-eliciting visuals have been studied extensively. Based on a number of previous studies on subliminal emotions, the Early Posterior Negativity (EPN) measured in the occipito-parietal area of the scalp was chosen as the event-related potential (ERP) of interest, as it has been reported to reflect the subliminal processing of faces, facial expressions, and other visuals of evolutionary interest such as dangerous animals. Three light directions, (1) silhouette light that completely hides facial features, (2) underlight that comes from below the face and distorts those features, and (3) toplight that hides the eyes, were found to elicit a statistically more negative EPN than 45-degree light, a lighting style that reveals the whole face, gives the subject depth and separation from the background, and is therefore often used as the chosen key light direction in filmmaking and portrait photography, for example in the so-called three-point lighting technique. Contributing to cognitive film studies, these results indicate that the way a character's face is lit affects the film experience as a whole already at the subliminal level of emotional processing.