Understanding Children’s Emotion-Specific Biases: How They Relate to Child-Level Factors, Emotion Recognition Accuracy, and Social Behaviour.

• In Event: New Directions in Studying Adult and Children’s Emotion Knowledge

Abstract

Introduction. Children’s ability to recognise emotions accurately has been associated with their social competence and academic achievement (Denham et al., 1990, 2002, 2014). Child-level factors like age, gender, verbal ability, ethnicity, race, maternal education, and dual-language learning have been shown to account for as much as 25% of the variance in children’s emotion recognition accuracy (Denham & Couchoud, 1990; Martin et al., 2010; McClure, 2000; Pons et al., 2004; Schultz et al., 2000; Widen & Russell, 2002). However, children’s emotion recognition biases (i.e., their tendency to over-identify emotions in the lack of corresponding cues) may explain additional variance. Some evidence supports this, but studies have primarily been conducted with clinical populations without controlling for child-level factors (Martin et al., 2010; Schultz et al., 2000). Methods. Patterns in children’s emotion-specific accuracies and biases, how they related to within-child factors, and how emotion-specific biases associated with accuracies and social behaviour were investigated in a community sample (N=876) of preschool-aged children. Results. Using structural equation modelling, results revealed that anger bias was the strongest bias with children incorrectly giving anger responses on 14% of non-anger items. Sad bias was the second strongest (12%), followed by happy (9%), fear (7%), and calm (4%). As expected, verbal ability consistently associated with all emotion-specific accuracies, βs=0.17–0.44, ps=<.001. Age associated with all accuracies, βs=0.22–0.28, ps=<.001, except for calm. Gender associated with happy, β=0.08, p=.037, and sad, βs=0.09, p=.030, but not anger, surprise, or calm. Except for maternal education associating with fear, β=0.13, p=.003, and speaking Spanish associating with calm, β=0.09, p=.004, there were no consistent associations with race and ethnicity. Similar to accuracies, all biases except calm were predicted by children’s age, βs=-0.12–-0.15, ps=<.001–.021. Unlike accuracies, verbal ability, gender, and other child level variables did not have any consisent associations with the biases. Altogether, more variance was explained among accuracies, R2s=.10–.33, than biases, R2s=.03–.06. After controlling for child-level factors, there were statistically significant small-to-moderate relationships between biases and accuracies, βs=0.07–0.26, ps=< .001–.048, but excluding fear bias, no bias was associated with same-emotion accuracy (e.g., anger bias and anger accuracy). Regarding children’s social behaviour, there was no specific pattern with biases. Anger and aggression and anxious and withdrawal were not statistically significantly associated with any emotion biases, and social competence was weakly associated with sad and calm biases, β=-0.09–-0.10, ps=.004–.012. Discussion. Findings are best explained using the Dual-Route Model of Emotion Recognition suggesting biases as a learned hursitic or “shortcut” that results in accuracy for some emotions but innaccury for others. The study highlights that non-clinical children also display emotion recognition biases as high as 12% depending on the emotion.

Authors

• Anushay Mazhar, Yale University
  Presenting Author

• Craig S. Bailey, Yale School of Medicine