David RandKey Mechanisms
Universal Drivers of Belief (regardless of truth value)
Repetition: Repeated exposure to a claim increases perceived truthfulness even when the claim is outlandish and politically opposed. This is a low-level cognitive effect; even cognitively sophisticated people are vulnerable. Rand's research on 2016 election misinformation found that seeing false claims multiple times increased belief in them, contrary to motivated-reasoning predictions.
Consistency with prior beliefs: People are more likely to believe claims aligned with their political views or factual understanding. This is often framed as "motivated reasoning"—partisan bias capturing cognition. But Rand argues a simpler, more rational explanation: if someone has different information sources (e.g., Fox News vs. MSNBC), they naturally have different factual understandings. Disbelieving information that contradicts your knowledge is actually reasonable when source reliability is uncertain.
Trusted sources: Endorsements from political elites and trusted figures increase belief. This is also rational: if you generally agree with a leader on many issues, using their position as a heuristic on unfamiliar topics is reasonable inference. However, elites' platforms and trust create asymmetries: followers encounter elite-endorsed false claims repeatedly but rarely see corrective evidence, making coordinated misinformation campaigns by elites particularly powerful.
Selective Vulnerability to Falsehoods
Lack of critical thinking / cognitive reflection: People who engage in more intuitive, "go with your gut" thinking are more vulnerable to false claims regardless of political alignment. This is distinct from partisan bias. Cross-cultural research across 16 countries on six continents consistently shows that cognitive reflection—the inclination to pause and think through a problem—reduces false-content belief. Importantly, if prior beliefs are already corrupted (e.g., down the "rabbit hole"), more reasoning just entrenches false beliefs through motivated reasoning.
Low digital literacy: Familiarity with internet concepts, knowledge of how Facebook's news feed works, and understanding of how news is produced all independently predict resistance to misinformation — separate from general education or critical thinking ability. Digital literacy trainings can help, though not all trainings are equally effective (testing is essential).
Distraction: On social media, news mixed with entertainment reduces critical thinking. Emotional activation also increases vulnerability to falsehoods.
Evidence and Interventions
1. Corrections and Debunking Work
Contrary to the backfire-effect hypothesis (corrections make people believe falsehoods more), Rand's research finds that warnings, labels, and corrections do reduce false belief. The effect is general and reliable across corrections' presentation details; the main challenge is scale—there are too many false claims to fact-check and too few corrective messages reach people who need them.
2. Crowdsourced Fact-Checking and Wisdom of Crowds
Lay evaluators can match professional fact-checker accuracy when their judgments are aggregated. Crowds of 20 or fewer people rating headlines generate agreement with professional fact-checkers at rates equal to inter-rater agreement among the professionals themselves. This approach has been validated across cultures, offering a scalable alternative to expert-only fact-checking. Platforms can use lay signals to label or de-amplify potentially problematic content.
3. Accuracy Nudges
Research shows most people don't want to share misinformation, but on social media they forget to consider accuracy because attention is directed to other signals (likes, shares, emotional arousal). Simple nudges—asking people to rate the accuracy of one random post, providing minimal digital literacy tips, or asking "How accurate is this?"—activate the concept of accuracy and cause people to share more truthful content. This works across countries and in large-scale Twitter field experiments, with 2–10% improvements in content quality.
Critical Mechanisms
Lazy vs. biased: Misinformation spread is driven less by partisan motivated reasoning than by lazy thinking—fast, intuitive judgments without deliberation. This has implications: interventions should target attention, not re-motivate reasoning (which can entrench false beliefs if priors are corrupted).
Exposure, not integrity: Observed differences in misinformation sharing across political groups reflect what people are exposed to, not differences in willingness to spread falsehoods. When people encounter false claims (because media diets differ), they share them at similar rates.
The scalability problem: Interventions must work at platform scale. A five-minute training video is useless if nobody watches it. Nudges succeed because they require <1 second of attention.
Broader Implications
Political elites and mainstream media—not just social media—play major roles in creating and spreading misinformation. These traditional sources often escape scrutiny despite high reach and perceived credibility.
Positive framing: It is not a post-truth world where facts don't matter. People do care about accuracy. The problem is exposure: people are not getting good information, particularly in certain media ecosystems. Increasing exposure to accurate information is a key leverage point.
Notes
This talk distills the psychological foundations that underpin Rand's later work on interventions. It succeeds in making mechanisms accessible: repetition, consistency, and trusted sources are intuitive levers; the insight that lack of reasoning (not bias) drives false sharing reframes mitigation strategies away from counter-persuasion toward attention design.
The cross-cultural evidence on cognitive reflection is particularly strong—an unusual finding that cognitive factors outweigh partisan ones. The talk also clearly separates rational inference from motivated reasoning, a distinction often lost in popular accounts of misinformation.
Key limitations: the talk does not deeply address content moderation dilemmas (who decides what is false), the difficulty of removing false beliefs once entrenched, or the increasing role of generative AI in misinformation production. It also largely skips the supply-side (incentives for producing falsehoods) in favor of the demand-side (why people believe/share them).