A database of naturalistic expressive faces for studying high arousal states

Abstract

Recent studies comparing behavior to different sets of facial stimuli have highlighted a need to employ more naturalistic, genuine facial expressions in cognitive research. To address this need, we identified and selected a large set of highly expressive face stimuli from the public domain, and used these stimuli to test whether participants can recognise when others are experiencing pain from their facial behaviour. After identifying 315 expressive faces to represent the kinds of facial behaviours often seen in three distinct contexts (i.e., injury-related, loss-related and victory-related), we ran six behavioural ratings tasks to characterise these faces along six dimensions; level of arousal, emotional valence, level of physical pain, attractiveness, familiarity, and perceived gender. The results indicate that injury-related expressions are recognised as lower in emotional valence than victory-related expressions, and higher in psychological arousal than both victory- and loss-related expressions. Overall, these findings suggest that the intense, energetic expressions of people in competitive situations are not rendered ambiguous to third parties by increased arousal. These results validate the use of naturalistic facial expressions in studies of non-verbal, injury-related behaviours and their recognition in forensic and clinical settings.

Perspective

Here we created and validated a large set of visual stimuli, which have been made available to the scientific community. Our results demonstrate that among high-arousal states, expressions related to feelings of intense pain and injury are visually distinct from expressions related to loss or triumph. Thus, the Wild Faces Database – High Arousal States (WFD-HAS) extension provides an important tool for understanding how we recognise injury-related facial expressions in the real world.