Using Facial Expressions, Vocalizations, and Postures to Characterize Children’s Emotional Responses to Corrective Feedback

• In Event: Beyond Self-Report Measures: Unlocking Insights Into Cognitive and Affective Feedback Processes With Multimodal Data Sources

Abstract

Objectives
Corrective feedback provides information about the learner’s progress on a task. Though generally helpful, this feedback can evoke different emotions. Few studies have directly assessed learner’s in-the-moment emotional responses during problem solving. We conducted an empirical study with elementary school children to quantify their emotional responses to corrective feedback. Rather than relying on self-report, we directly coded their affect based on visible behaviors.

Theoretical Framework
Multiple theories suggest that feedback can produce affective responses that interfere with task performance. For example, Feedback Intervention Theory (Kluger & DeNisi, 1996) suggests that feedback can direct attention toward the self (e.g., “I got that wrong I must not be smart”) in ways that produce negative emotions, which in turn consume cognitive resources. More recently, the Model of Motivated Feedback Disengagement (Grundmann et al., 2021) suggests that these negative emotions increase the salience of hedonic goals and can push learners to disengage with the task to avoid feeling bad. We tested these theoretical assumptions in children.

Methods
Participants included 109 elementary school children (Mage=7.4 years, SD=0.9; 44% female; 65% white) from a midwestern community in the U.S.. Each child completed a study session via Zoom or in the lab with a trained researcher. They solved a series of math equivalence problems (e.g., 6+4+8=3+__). There were five sets of four problems each, and children could choose to stop after each set. Children received trial-by-trial corrective feedback that included an explicit right/wrong judgment and the correct answer.

Data Sources
We recorded children’s accuracy on the math problems and their persistence (when they chose to stop). Also, trained researchers watched the recorded videos of these sessions and rated children’s affect. For each instance of feedback, the researcher rated the child’s positive affect on a scale from 1 (not-characteristic) to 5 (exceptionally-characteristic) and also rated their negative affect on a similar scale. The researcher relied on facial expressions (e.g., smiles, frowns), posture (e.g., slumping), and tone of voice.

Results
The average accuracy was 64% (SD=33%), and average persistence was choosing to complete 13.3 items (out of 20; SD=6.2). Children were modest in their expression of emotion. Across all trials, average positive affect was 1.57 (scale from 1 to 5; SD=.53) and average negative affect was 1.35 (SD=.50). Negative emotion, but not positive emotion, was related to children’s engagement. Positive affect was not correlated with persistence, r(107)=.01, p=.914, but negative affect was negatively correlated with persistence, r(107)=-.28, p=.004. Also, 40 children opted to complete all 20 math problems and they experienced less negative emotion (M=1.23, SD=0.50) relative to the 69 children who ended the task early (M=1.43, SD=0.49), F(1, 107)=3.98, p=.04.

Scholarly Significance
Our study suggests that children’s behaviors are a reliable indicator of their emotional responses to feedback during problem solving. Though most children did not experience extreme emotional episodes, their modest experiences of negative emotion were related to stopping the task early. These empirical findings support theoretical assumptions about the relationship between negative affect and task disengagement (Grundmann et al., 2021).

Authors

• Emily R. Fyfe, Indiana University

• Megan Merrick, Indiana University