Gordon PennycookContributions
- Challenges political-partisanship narrative: Shows people are actually better at discerning truth from falsehood for politically concordant news, contradicting the popular belief that politics is the primary driver of fake news belief.
- Identifies reasoning as central mechanism: Demonstrates that more reflective, analytically-minded people are better at distinguishing true from false content across ideological boundaries.
- Maps heuristic influences: Documents how familiar content, source credibility, emotional evocation, and processing fluency drive misinformation belief through cognitive shortcuts.
- Quantifies belief-sharing disconnect: Establishes that social media sharing of false content is only weakly correlated with actual belief in that content; inattention accounts for roughly half of false-news sharing.
- Evaluates interventions comprehensively: Reviews fact-checking, warnings, inoculation approaches, and accuracy-salience prompts, identifying scalable solutions and fundamental limitations of each.
Method
This is a literature review synthesizing recent empirical work on misinformation psychology. The paper covers 14+ multi-study datasets examining truth discernment across U.S. samples and meta-analytic findings on reasoning, partisanship, heuristics, and sharing behavior. Key analyses include pooled studies on cognitive reflection and truth discernment (Figure 1) and large-scale Twitter experiments on accuracy prompts and publisher-credibility crowdsourcing (Figures 2–3).
Results
Truth discernment and reasoning: Across 14 studies, people who scored higher on cognitive reflection (a measure of analytical thinking) were significantly better at distinguishing true from false news, regardless of political concordance. The effect of reasoning on accuracy was roughly twice as large as the effect of political concordance. More reflective participants also showed better discernment when evaluating politically discordant headlines.
Belief versus sharing: When participants were asked to rate headline accuracy and then decide whether to share them, accuracy judgments predicted sharing only weakly. In a baseline condition, 91% more sharing occurred for false headlines than accuracy judgments alone would predict. Three mechanisms explain the disconnect: confusion-based (33% of sharing when accuracy is explicitly checked beforehand), preference-based (16% willing to share despite knowing falsity), and inattention-based (51% reduced when accuracy was salient—inattention is responsible for roughly half).
Heuristic cues: Familiarity, source credibility, emotional evocation, and social feedback (e.g., likes) all drive misinformation belief independently of political ideology.
Interventions: Accuracy prompts deployed on Twitter significantly increased the quality of content subsequently shared; crowdsourced accuracy ratings from laypeople correlated highly with professional fact-checkers; asking users to explain how they know a headline's truth value before sharing increases discernment. Limitations: fact-checking is not scalable; corrections can backfire under high uncertainty; warnings may create an "implied truth" effect for unmarked content.
Notes
A foundational synthesis that reframes the misinformation problem away from motivated reasoning toward inattention and limited cognitive resources. The paper's key insight—that people are often unintentionally poor truth-discriminators rather than intentionally deceitful—shifts the design space for interventions toward cognitive aids rather than persuasion. The distinction between belief and sharing has important implications for platform interventions: social media may amplify misinformation not primarily because people are confused, but because sharing decisions are made in an attention-poor context. The accuracy-prompt work is particularly strong and has influenced platform policy (e.g., Twitter, Facebook).