Pennycook, G., & Rand, D. G. (2021). The psychology of fake news. Trends Cogn. Sci., 25, 388–402. https://doi.org/10.31234/osf.io/ar96c
Summary
This review by Pennycook and Rand synthesizes psychological research on why people believe and share false news online, and mounts a sustained challenge to the dominant narrative that partisanship and politically motivated reasoning are the chief culprits. Drawing on cognitive science and dual-process theory, the authors argue that poor truth discernment stems primarily from a failure to engage careful, reflective reasoning, from insufficient or inaccurate prior knowledge, and from reliance on heuristics such as familiarity. They further document a striking gap between what people believe and what they share, attributing it largely to inattention rather than confusion or deliberate misinformation-spreading. The paper closes by consolidating evidence on scalable interventions—accuracy nudges and crowdsourced veracity ratings in particular.
Key Contributions
- Reframes fake-news susceptibility from a politically-motivated-reasoning account toward an inattention-and-reasoning (“classical reasoning”) account.
- Clarifies the conceptual distinction between truth discernment (belief in true minus false news, akin to signal-detection d′) and overall belief (bias), showing why factors that move belief need not affect discernment.
- Identifies the belief–sharing dissociation and adjudicates among confusion-based, preference-based, and inattention-based explanations, favoring inattention.
- Consolidates evidence on the effectiveness and limits of anti-misinformation interventions, especially scalable accuracy prompts and crowdsourced ratings.
- Positions online misinformation as an ecologically valid domain uniting lab experiments, large-scale social media data, and field experiments.
Methods
A narrative synthesis of cognitive, social, and political psychology research, framed through signal detection theory and System 1/System 2 dual-process models (using instruments such as the Cognitive Reflection Test and Berlin Numeracy Test). The authors pool and reanalyze data from 14 US studies rating true/false, Democrat- and Republican-consistent headlines, comparing high vs. low reflective reasoners and concordant vs. discordant content. They also draw on survey experiments contrasting accuracy-rating and sharing-intention conditions, a large-scale Twitter field experiment sending accuracy prompts, and crowdsourcing studies benchmarking layperson ratings against professional fact-checkers (NewsGuard).
Findings
- A headline’s actual veracity affects belief far more than political concordance—“politics does not trump truth”—and people are somewhat better at discernment for concordant news even while believing it more overall.
- More reflective individuals show better truth discernment regardless of concordance; in the pooled data the reasoning-based difference in discernment was roughly twice the partisanship-based difference.
- Experimentally increasing deliberation reduces belief in false (but not true) news without increasing polarization; overconfidence and reliance on emotion increase susceptibility.
- Familiarity (illusory truth), source/elite cues, social feedback, and emotional content act as heuristics that inflate belief in false claims.
- Sharing intentions are largely insensitive to veracity; in one study, false-headline sharing intentions were 91% higher than accuracy assessments, and an accuracy prompt cut false sharing by 51%—implicating inattention.
- Automated detection and professional fact-checking face limits (subjectivity, non-scalability, implied-truth effects), while accuracy prompts, literacy tips, prebunking, and crowdsourced ratings show promise; layperson crowds correlate highly with professionals (r ≈ 0.90 for publisher trust).
Connections
This review is a foundational statement of the inattention-and-reasoning paradigm that many later intervention studies build on, particularly work on accuracy nudges and crowdsourced veracity assessment such as Mosleh2024-op and the debunking-focused Costello2024-bg. Its critique of motivated-reasoning accounts speaks directly to partisanship-centered analyses of misinformation sharing like Osmundsen2021-et, while its emphasis on prebunking/inoculation connects to the psychological-resistance line pursued in van-der-Linden2026-jt. The finding that fake news occupies a small share of media diets is complemented by exposure-scale work including Allen2020-nj, Grinberg2019-ua, and Guess2019-ym, and its treatment of viral spread relates to Vosoughi2018-at.