Investigating epistemic emotions experienced while reading refutation texts through a fine-grained measure of emotion

Abstract

The current study addressed the often-overlooked role of epistemic emotions in refuting misinformation by replicating and expanding on the work of Trevors and Kendeou (2020). It broadened the participant pool beyond well-educated college students and introduced a novel dynamic measure, "DynamicEmo", to capture epistemic emotions experienced while reading refutation texts in a more fine-grained way. Results reaffirmed that positive, negative, and standard refutation texts (vs. non-refutation texts) effectively enhanced knowledge revision. Analysis using DynamicEmo revealed that paragraphs presenting inconsistent information (misinformation+correction) in refutation texts elicited activating (curiosity and confusion) or suppressed deactivating epistemic emotions (boredom). Notably, in-the-moment negative epistemic emotions, triggered by critical correct-outcome sentences, were negatively predictive of knowledge revision, highlighting the significance of emotions experienced during critical parts of refutation text reading. This study demonstrated the key role of epistemic emotions in knowledge revision, while offering more granular insights through dynamic emotion measurement compared to traditional post-hoc self-reports.

Fig. 1. The DynamicEmo measure.

This instrument enables participants to report emotions experienced while reading each sentence, thus allowing for granular data analysis. Also, multiple emotions and their dynamics can be captured within a single text.

Fig. 2. Measures of emotion in reading.

These measures are assessed based on types (experiential, physiological, and behavioral), dynamics (online and offline methods), and considerations of cost and accessibility.

Fig. 3. Descriptive analyses of knowledge posttest results and epistemic emotions.

Bar graph showing means and 95% confidence intervals for a knowledge post-test scores and b counts of epistemic emotions measured through DynamicEmo across different text conditions.

Fig. 4. Predicted probability of postExplanation by text conditions and emotions.

The plot illustrates the estimated probability of knowledge posttest scores as a function of experienced emotions toward the correct-outcome sentence under different text conditions.