Investigating the effects of an educational infographic and cognitive load on deepfake detection and metacognitive judgments

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

This article is not in any list yet, why not save it to one of your lists.
Log in to save this article

Abstract

AI-generated deepfake videos, which mimic real recordings of individuals, enable the impersonation of public figures and the widespread dissemination of misinformation. According to previous research, people are not only poor at detecting deepfakes but also often overestimate their detection ability, highlighting the need for effective interventions. Despite this, evidence on how to improve deepfake detection and calibrate metacognitive judgments remains limited. In this preregistered 2 x 2 experiment, we examined the effects of an educational infographic, cognitive load, and their interaction on deepfake detection accuracy and metacognitive overestimation (absolute and relative). We also investigated age, cognitive reflection, and general trust as potential moderators. A total of 229 participants were randomly assigned to infographic versus control conditions and high versus low cognitive load conditions, viewed 12 political videos (deepfakes and authentic videos), and rated videos’ authenticity and their perceived performance. The results showed that participants exposed to the infographic showed higher detection accuracy and more accurate relative performance judgments, whereas cognitive load did not produce significant main effects. An interaction suggested that the infographic’s metacognitive benefits weakened under higher load, although this finding requires caution due to an inconclusive manipulation check. None of the individual differences moderated the findings. Lastly, participants slightly underestimated their absolute performance but considerably overestimated their relative standing, with the lowest performers showing the strongest overconfidence. Overall, the results suggest that concise and engaging guidance, such as infographics, can enhance deepfake detection and promote more calibrated self-assessments, providing a scalable tool for strengthening resilience to synthetic media.

Article activity feed