Category: Child / Adolescent - Anxiety
Emotional coherence refers to links among an individual’s emotional responses across subjective, behavioral, and physiological systems (Mauss et al., 2015). Low coherence among systems is a robust risk factor for psychopathology. One barrier to studying emotional coherence is the extensive expertise needed to assess and interpret data across systems. Automatic emotion coding programs may facilitate the study of emotional coherence and its links to psychopathology. We tested one program, IntraFace, which consists of free and easy-to-use software for coding facial expressions from videotaped observations using Ekman and colleagues’ (2002) FACS paradigm for identifying specific emotions (i.e., happy, surprised, neutral, disgust, sad; Torre et al., 2015).
We recruited 60, 14-15 year-old adolescents (30 clinic-referred; 30 community control) with groups matched on age and gender. Adolescents participated in counterbalanced social interaction tasks with peer confederates consisting of a structured social interaction task (SSIT; Curran, 1982), unstructured conversation task (UCT; Turner et al., 1994), and impromptu speech task (IST; Beidel et al., 2010). To study emotional coherence, we used IntraFace to process video-tape recordings of these tasks and identify anxiety-relevant emotions (i.e. probability estimates of “surprised”; range: 0-10). After each interaction, adolescents self-reported their arousal using the Self-Assessment Manikin (SAM; Bradley & Lang, 1994).
A repeated-measures ANOVA tested differences across tasks (SSIT, UCT, IST) and groups (clinic-referred, community control) in probabilities of displaying the emotion surprised. There was a significant task X group quadratic effect (F=7.26; p < .01). Clinic-referred adolescents showed greater probabilities for displaying surprised during unstructured tasks (UCT: M=3.86) relative to structured tasks (SSIT: M=3.02; IST: M=3.09), whereas community control adolescents showed greater probabilities for displaying surprised during structured tasks (SSIT: M=3.45; IST: M=3.66) relative to unstructured tasks (UCT: M=3.16). In contrast, for adolescents’ SAM ratings, there was a non-significant task X group quadratic effect (p=.09), a significant linear task effect (F=153.395, p < .001), and a significant group effect (F=5.70, p < .05). Although both groups self-reported their highest levels of arousal for the IST, followed by the UCT, and then the SSIT, clinic-referred adolescents self-reported significantly greater levels of arousal across tasks relative to community control adolescents.
IntraFace facilitates studying emotional coherence by yielding data that complement alternative emotion modalities (i.e., self-reported arousal), and significantly distinguish clinic-referred from non-referred adolescents.
Michelle Truong– Undergraduate Research Coordinator, University of Maryland at College Park
Erica Rausch– Undergraduate Research Coordinator, University of Maryland at College Park
Lauren Keeley– Full-Time Research Coordinator, University of Maryland at College Park, Adelphi, Maryland
James Riffle– Research Coordinator, University of Maryland at College Park
Sebastian Szollos– Research Coordinator, The University of Maryland, Columbia, Maryland
Bridget Makol– Graduate Student, University of Maryland, College Park, Chicago, Illinois
Tara Augenstein– Graduate Student, The University of Maryland
Melanie Lipton– Graduate Student, The University of Maryland
Sarah Racz– Clinical Assistant Professor, University of Maryland at College Park
Andres De Los Reyes– Associate Professor, University of Maryland at College Park, College Park, Maryland
Undergraduate Research Coordinator
University of Maryland at College Park
Full-Time Research Coordinator
University of Maryland at College Park