Correction Effects on the Perceived Accuracy of Fake News Headlines Over Time: The Critical Role of Recollecting Corrections

Fake news exposure on social media can diminish the accuracy of memory and beliefs. To mitigate these effects, social media sites issue fact-checked corrections. Competing theories propose that repeating fake news in corrections will either increase false beliefs by enhancing familiarity (familiarity backfire account) or reduce false beliefs by increasing co-activation and integration of real and fake details (integration account). Reductions in fake news beliefs also depend on the degree to which corrections promote recollection of their occurrence. In two experiments, we tested these accounts and examined the role of correction recollection in reducing the perceived accuracy of fake news headlines from the internet. Participants first rated the familiarity and accuracy of real and fake news of unclear veracity. Next, they read veracity-labeled corrections of fake news with only fake news details, only real news details, or fake before real news details. Finally, they re-rated the accuracy of real and fake news and rated their memory for corrections immediately and after a week (Experiment 1) or month (Experiment 2). On immediate tests, corrections with fake news, with and without real news, reduced accuracy ratings for fake news more than corrections with only real news. Fake news accuracy ratings regressed over time and remained lowest for corrections with both fake and real news. Such reductions depended on correction recollection, which occurred most for corrections with fake and real news. These findings are compatible with the view that including fake and real news in corrections promotes integration and subsequent recollection of corrections.