Emotional disorder symptoms and emotion regulation by cognitive reappraisal in individuals with a history of psychopathology: A multi-method approach

Introduction: Internalizing symptoms are characterized by difficulty regulating negative emotions and associated with excessive self-referential focus, cognitive inflexibility, and avoidant coping. Cognitive reappraisal, the process of reinterpreting negative emotion stimuli, is the most widely studied and effective emotion regulation (ER) strategy. Natural language processing (NLP) metrics have emerged as a way of understanding emotions and could also help understand ER via reappraisal. We explored the combination of theory-relevant NLP metrics indexing emotionality, self-referential focus, cognitive inflexibility (“certainty”), and avoidant coping (“action”) to understand (1) ER during reappraisal, (2) individual differences in ER, and (3) shorter, 2-week, and longer-term, 8-week, internalizing symptoms. Methods: Individuals with a history of psychopathology (N=409) participated in a randomized controlled trial. We analyzed text and self-report data from the cognitive restructuring portion of the interventions. The text was analyzed with Sentiment Analysis aNd social Cognition Engine (SEANCE), which pools together scores from various NLP algorithms. Results: Following reappraisal, there were large reductions in self-references and negative language with smaller increases in action and certainty. The SEANCE metrics did not appear to correlate with self-reported improvements during the task. However, self-reported reappraisal was associated with larger changes in SEANCE-certainty during the task. These changes in SEANCE-certainty also showed predictive validity in relation to well-being and internalizing symptoms in the shorter-term symptoms. Discussion: NLP metrics may capture individual differences in reappraisal, and some, but not all, aspects of the process. Consistent with theory, changes in NLP-measured certainty may capture an important aspect of reappraisal.