Exploring how political deepfakes trigger cognitive processes and downstream psychological and behavioral outcomes: A systematic review

This review examines the cognitive and psychological effects, and behavioral implications of political deepfakes. Drawing on the PRISMA 2020 guidelines, this study synthesizes 25 empirical studies. Across the studies, political deepfakes are found to have no higher inherent persuasive effect over audiences compared to other misinformation, but they produce distinctive effects through epistemic destabilization. As a result, trust erosion emerges as a crucial outcome of exposure to deepfakes. However, the effects are highly context-dependent, with deepfakes producing substantial effects in polarized democracies. Individual factors, including cognitive ability, partisan identity, and motivated reasoning, systematically moderate responses. Moreover, sharing of deepfakes, a major behavioral outcome, relies on identity signaling and partisan alignment rather than credibility judgments. Measures to contain the implications of deepfakes, like issuing warnings, can backfire, resulting in paradoxical outcomes like increased generalized distrust. Epistemic dominance was evident across the studies, with authors from three countries (the US, the Netherlands, and the UK), accounting for nearly 50% of the authorship. Moreover, this study identifies systematic context-dependency, measurement inconsistencies, and the deepfakes paradox. Therefore, this paper provides crucial information for future studies, specifically on developing valid scales, conducting longitudinal studies, and conducting studies in non-Western and low-polarized contexts. Finally, it has significance for policy formulation to prevent misuse of deepfakes where the state of democracy, levels of political polarization, population characteristics, and behavioral tendencies inclination should be carefully considered.