Evaluation of the Presence of a Face Search Advantage in Chernoff Faces

Chernoff faces (Chernoff, 1973) are an early attempt at large scale multivariate data representation and are based on underlying assumptions that have not been empirically tested. This study investigated i) whether data coded as Chernoff faces benefit from face perception, and ii) whether each feature of a Chernoff face is equally salient. We tested four pairs of oppositely coded Chernoff faces (e.g. smile, frown) in an oddball search paradigm with set sizes of 5, 10 and 15. To evaluate whether face perception aided search, we used a control condition with inverted faces, a manipulation known to diminish holistic face processing. Equivalent search efficiencies for upright and inverted Chernoff faces demonstrates that they do not receive any significant benefit from face perception. Additionally, none of the features tested together produced significantly different search efficiencies from one another. It also appears that overall Chernoff faces do not allow for particularly efficient visual search.