Knowledge that Transforms

Abstract People typically read the titles of content (e.g., an article or post) before deciding whether to consume it. Titles can be composed as a statement (e.g., “Green Tea Has Health Benefits”) or a question (e.g., “Does Green Tea Have Health Benefits?”). Question titles are common and are believed by journalists to increase engagement. The present research examines whether this is the case. Four studies—including archival analyses of Reddit posts ( N = 53,030) and academic articles ( N = 3,078,791), online news headline experiments ( N = 22,743 A/B experiments), and a pre‐registered lab study ( N = 400)—indicate that question‐framed titles reduce reader engagement because they are viewed as less informative. These findings suggest that those wishing to maximize engagement with written content should avoid using questions in their titles.

Abstract Despite longstanding assumptions that embarrassment leads to avoidance, empirical findings remain inconsistent. This meta‐analysis synthesizes 152 independent samples ( N = 64,374) to clarify when and how embarrassment motivates avoidance. Drawing on Protection Motivation Theory, we find a curvilinear effect, where avoidance is strongest in moderately intense, embarrassing situations, but diminishes when embarrassment is either too weak to prompt concern or so intense that the consequences of avoidance become too severe to risk. This pattern underscores that avoidance is most likely when individuals perceive a moderate threat and believe avoidance is effective. Moderator analyses reveal that this effect is stronger for product‐related vs. medical contexts, when others are present, when embarrassment is anticipated (vs. experienced), and in collectivist cultures. Comparisons with guilt and shame further demonstrate that embarrassment operates through distinct, socially driven appraisals. These findings reconcile inconsistent findings in the embarrassment literature and offer guidance for reducing harmful avoidance by aligning threat and coping appraisals to promote engagement with socially sensitive behaviors. Implications for theory and practice, as well as future research priorities, are discussed.

Abstract A common assumption among scientists, journalists, and policymakers is that combating misinformation reliably changes behavior. However, the empirical evidence reveals that the belief–behavior association is often modest, variable, and context‐dependent, raising critical questions about when and how to pursue belief versus behavior change. In this paper, we discuss the mechanisms by which beliefs influence behavior and the conditions under which addressing beliefs can change behavior. Specifically, we review the belief‐to‐behavior inference model, which proposes that beliefs influence behavior when (a) the belief is linked to a behavioral goal, (b) the inferential path from belief to behavior is relatively short, and (c) the belief–behavior association is preserved in memory. Our framework aligns intervention decisions with the cognitive architecture of belief–behavior correspondence and the intervention's goals, whether maximizing belief accuracy or behavioral impact. We also reviewed individual and social‐structural interventions that are best suited for changing beliefs versus behavior, conceptually integrating interdisciplinary work on behavior change with the psychology of belief change.