Cognitive Load, Emotional Asymmetry, and the Third-Person Effect: Explaining Audience Responses to AI-Generated Health Misinformation

This study extends the Third-Person Effect (TPE) framework into the era of generative artificial intelligence by examining how cognitive and emotional mechanisms shape audience responses to AI-generated healthmisinformation. Across a two-stage online experiment, participants were exposed to AI- or human-writtenhealth messages and later received one of three literacy interventions (AI principle education, fact-checking, or emotional inoculation). Results showed that higher AI literacy, critical thinking, and verification efficacy amplified Third-Person Perception (TPP), whereas cognitive load weakened it, revealing an “ability–difficulty” moderation. Negative emotions such as fear, anxiety, and confusion increased TPP, while hope reduced it. Literacy interventions, particularly emotional inoculation, significantly decreased TPP and enhanced detection accuracy, with route–audience congruence effects consistent with the Elaboration Likelihood Model. These findings advance TPE theory beyond anthropocentric media contexts, conceptualizing it as a technocognitive bias moderated by cognitive resources and emotional asymmetry. The study further demonstrates that targeted literacy interventions can recalibrate public perception and trust toward AI-mediated health communication.