Abstract
Throughout man's time on Earth, one of the most consistent threats to his chances of survival has been other humans. It would seem, therefore, evolutionarily expedient to provide humans with every opportunity to detect the threat posed by individuals around them. Previous thinking along these lines has prompted some researchers to argue that certain facial expressions might be subject to enhanced processing to maximize the speed and accuracy with which humans locate individuals bearing threatening expressions. Evidence supporting this proposal comes largely from visual search tasks which have demonstrated that threatening expressions are more rapidly detected than non-threatening ones. An open criticism of this effect is that it may be due to low-level visual artifacts, rather than biological preparedness. One successful approach for controlling low-level, image-based differences has been to use schematic faces (simplified line drawings). We report experiments aimed at discovering whether the enhanced processing of threatening schematic faces, might also be due to low-level features within the stimuli. The first study replicated the standard threat search advantage, but also measured an effect using similar stimuli comprised of obliquely oriented lines. The effect was also present with these stimuli rotated, a manipulation which served to remove any residual resemblance the abstract images had to a face. The results suggest that low-level features underlie the search advantage for angry, schematic faces, thereby undermining a key source of evidence of a search advantage for specific facial expressions. As an interesting aside: whatever the features happen to be, that are responsible for this effect, they are not captured in simple saliency models such as that of Itti & Koch (Vision Research, 2001).
Australian Research Council.