Misinformation: susceptibility, spread, and interventions to immunize the public

Sander van der Linden

doi:10.1038/s41591-022-01713-6

In brief

This comprehensive review synthesizes the psychology of misinformation through three integrated dimensions: susceptibility (why people believe false claims), spread (how misinformation propagates in social networks), and immunization (therapeutic and prophylactic interventions to build resistance). The paper argues that understanding misinformation requires integrating accuracy-driven accounts of cognition with identity-protective motivated reasoning, that exposure does not equal persuasion on social media, and that preventive "prebunking" via psychological inoculation is more scalable than reactive debunking.

Contributions

Theoretical integration of susceptibility accounts: Reconciles the "inattention" account (people lack reasoning) and "motivated reasoning" account (people prioritize identity), demonstrating both operate in different contexts and suggesting future research needs contextual theories accounting for trade-offs between accuracy and social motives.
Epidemiological framing of spread: Applies SIR (susceptible-infected-recovered) models from epidemiology to quantify misinformation reproduction (R₀) in networks; documents that false news spreads faster on Twitter but notes estimates are platform-dependent and limited by narrow fact-check definitions.
Distinction between exposure and infection: Establishes that exposure to misinformation does not automatically lead to belief or sharing; people share for reasons beyond accuracy (e.g., "interesting if true"), requiring more nuanced models of spreading.
Synthesis of debunking best practices: Reviews meta-analyses on fact-checking effectiveness, finding that debunking works when designed to (1) lead with facts, (2) appeal to scientific consensus, (3) explain why misinformation is wrong, and (4) provide a credible alternative explanation, but suffers from limited reach and "continued influence" effects.
Evidence base for psychological inoculation: Catalogs emerging research on prebunking interventions, including narrow-spectrum (fact-based) and broad-spectrum (technique-based) inoculation, gamified approaches (Bad News, GoViral!), and active versus passive variants; documents that inoculation confers protection even after exposure and allows for "booster shots" to sustain immunity over time.

Method

Literature review and conceptual synthesis. The paper reviews empirical research across three dimensions:

Susceptibility: Reviews cognitive mechanisms (illusory truth effect, processing fluency), individual differences (age, political orientation, numeracy, analytic thinking), and two competing theoretical accounts (inattention vs. motivated reasoning).
Spread: Synthesizes research on Twitter, Facebook, and other platforms using cascade analysis and network epidemiology; highlights the role of echo chambers, superspreaders, and the distinction between exposure estimates and actual persuasion effects.
Immunization: Synthesizes meta-analyses and randomized controlled trials on debunking interventions and inoculation approaches, contrasting reactive (post-exposure) with proactive (pre-exposure) strategies.

Results

Susceptibility

The illusory truth effect (repeated claims are judged as true) is robust across ages, ideologies, and plausibility levels. However, susceptibility varies:

Age: Older individuals are more susceptible to fake news (with exceptions in COVID-19 context).
Political orientation: Right-wing individuals more susceptible, but effects vary by culture.
Cognitive factors: Numeracy, analytic thinking, and education predict lower susceptibility; however, the "motivated numeracy" account suggests high-education conservatives may become more polarized on politically salient topics.

The paper reconciles competing theories: inattention (lack of reasoning) and motivated reasoning (deliberate identity protection) both operate, with dominance depending on context. Future research should examine how people trade off accuracy against social/identity goals.

Spread

False news on Twitter spreads ~70% faster and reaches 1,500 people in ~6 times shorter than true news. However:

Fake news represents a small portion of most people's media diet.
A small number of "superspreaders" and "superconsumers" account for majority of spread.
Echo chambers aid virality of misinformation but impede corrections.
Exposure ≠ infection: Exposure estimates from social media data often do not align with self-reported exposure; crucially, people share misinformation for reasons beyond accuracy (e.g., "interesting if true").

The paper emphasizes the need for better understanding of conditions under which repeated exposure leads to actual belief change, especially given studies using laboratory designs may not represent real-world persuasion.

Immunization

Debunking (therapeutic/post-hoc):

Meta-analyses confirm fact-checking and debunking interventions reduce belief in false claims, including on health topics.
Effectiveness is attenuated by (1) quality of the debunk, (2) time elapsed, (3) prior beliefs/ideologies.
Best-practice debunking involves: leading with facts, appealing to scientific consensus, explaining why misinformation is wrong, and providing an alternative explanation (see Figure 1 in paper).
Risk of backfire effects (correction strengthens belief in myth) exists but is rarer than previously thought; worldview backfire is not supported in recent replications for vaccination topics.
Limitations: only ~40% of target audience reaches corrections; people continue making inferences from false claims even after acknowledging correction ("continued influence").

Prebunking/Inoculation (prophylactic/pre-emptive):

Psychological inoculation theory: pre-exposing people to weakened misinformation coupled with strong refutations builds resistance.
Recent advances: field has moved from narrow-spectrum (fact-specific) to broad-spectrum (technique-based) inoculation; people inoculated against common tactics (fake experts, emotional manipulation, conspiracy framing) show cross-protection against novel misinformation.
Gamified interventions (Bad News, GoViral!) enable active inoculation (players generate own "antibodies") at scale; randomized trials show effects on identification, confidence, and reduced sharing.
Immunity wanes over time but can be maintained via "booster shots."
Prebunking outperforms debunking in head-to-head comparisons; psychological inoculation can protect even after exposure.

Notes

This review is the authoritative synthesis of the post-pandemic misinformation literature, emphasizing a shift from reactive (detecting/correcting) to proactive (building immunity) approaches. Van der Linden's integration of susceptibility, spread, and intervention within a single framework helps clarify why some phenomena in the literature appear contradictory: the same intervention (e.g., analytical priming) may reduce susceptibility to non-political misinformation but fail or backfire on politically salient claims, depending on whether inattention or motivated reasoning dominates the context.

The paper's emphasis on the "exposure ≠ infection" distinction is important: much research conflates exposure estimates with actual persuasion. Similarly, the distinction between broad-spectrum (technique-based) and narrow-spectrum (fact-based) inoculation addresses scalability—technique-based approaches can protect against entire classes of misinformation without bespoke interventions for each false claim.

The review also identifies important limitations in both theoretical accounts: motivated numeracy effects are less replicable than initially believed; accuracy nudges show mixed results; and the equivalence paradox makes it hard to distinguish genuine motivated reasoning from selective exposure or legitimate source discounting.