Stimulus characteristics influence emotion recognition performance

• In Event: Poster Session 05
  In Poster Session: PS 05 Topic - Social, Emotional, Personality

Abstract

Researchers in the field of emotional development have raised concerns about the nature of the stimuli used to assess emotion recognition (Barret et al., 2019). Participants are often asked to identify emotions from static, disembodied photographs of faces, raising questions about the ecological validity of findings. Moreover, the impact of discordance between the age of participants and the target stimuli on performance is often not considered. In addition, emotion recognition tasks typically use emotion words as response options, which can influence how emotional stimuli are processed and may conflate emotion understanding with variability in language skills (Cole et al., 2010). These considerations contribute to a lack of consensus about how stimulus characteristics influence emotion recognition performance across development.

To address these challenges, we assessed the influence of various stimulus characteristics on emotion recognition. The current study is the first to evaluate differences in emotion recognition performance using static and dynamic images, child and adult faces, and facial close-up and full body stimuli. Children (N=54, Mage=9.05, SDage=1.14) and adults (N=51, Mage=19.61, SDage=1.25) completed an emotion recognition task where they were shown a 1 s video (full-body or facial close-up) or a 1 s still, close-up photograph (adult or child face) of an actor displaying one of five emotions: happy, sad, angry, fearful, or neutral (Figure 1a-d) with racial diversity in the actors. We introduce a novel and open-source cartoon scale, the “Eugenies,” that reduces reliance on language abilities to assess how well participants detect the core feature of these emotions (Figure 1e). Following the presentation of the emotional stimuli, participants chose the Eugenie that they believed best represented the emotion shown.

Overall, participants demonstrated high accuracy in matching the actor’s emotion with the correct cartoon image (Maccuracy=94%, SDaccuracy=6%, range=24%) Adults had overall higher accuracy than children (b=0.552, X2(1)=3.450, p=.047) which is consistent with research showing that emotion recognition abilities improve throughout childhood, reaching adult performance around the age of 14 years (Kolb, Wilson, & Taylor, 1992). Next, results indicate that adults were less accurate in identifying the emotions portrayed by child actors as compared to adult actors (b=-1.566, X2(3)=35.545, p<.001), but children showed no difference in performance for child versus adult actors (p=.472). Children were better at recognizing adult dynamic faces than adult static faces (b=0.678, X2(3)=17.424, p<.001) and were marginally better at recognizing adult dynamic bodies than adult static faces (b=0.355, X2(3)=17.424, p=.075). In contrast, adults performed equally well in rating adult emotions, regardless of stimulus type (p=.620). Finally, preliminary analyses examining emotion-specific differences suggest that the recognition of fear may be enhanced by the use of dynamic stimuli.

Taken together, these findings suggest that children’s emotion recognition performance may suffer when only static cues are available, supporting the view that the inclusion of dynamic cues in emotion recognition tasks likely increases task reliability and improves ecological validity. Findings will be discussed in the context of methodological influences on emotion recognition performance, implications for design and study planning, and future directions for emotion development research.

Authors

• Tralucia Powell, Boston University
  Presenting Author

• Rista Plate, University of Pennsylvania
  Non-Presenting Author

• Nicholas James Wagner, Boston University
  Non-Presenting Author

• Rebecca Waller, University of Pennsylvania
  Non-Presenting Author