Key themes
- The continued influence effect. Even when people encode, believe, and remember corrections, they continue to rely on the initial misinformation in their reasoning and memory. This is measured through inference tasks (open-ended questions, rating scales, behavioral intentions) rather than memory tests, demonstrating that the effect is cognitive (how people use information to reason) rather than memorial (what they remember).
- Source credibility and correction effectiveness. Experimental studies show that trustworthiness of the correction source matters significantly for misinformation reliance, but expertise does not—a replication of earlier work (Giloy & Garach, 2013). Participants are credulous to low-expertise high-trust sources and skeptical of high-expertise low-trust sources. This holds even when people strongly believe the correction itself.
- Worldview effects and political asymmetry. Initial Australian studies found that corrections of worldview-congruent misinformation were less effective for conservative participants—suggesting a political asymmetry in motivated reasoning. However, follow-up studies in the US with larger samples found no such asymmetry: retractions worked equally well across partisan lines. Recent literature shows the "backfire effect" and asymmetry claims are increasingly failing to replicate, though real-world data still show asymmetries in science denial that laboratory studies struggle to explain.
- Social norms and belief expression. Communicating to people that peer consensus has shifted (e.g., "only 37 of 100 peers endorse this claim") produces small but measurable reductions in stated belief, especially when norms suggest consensus against the claim. The effect is amplified in public contexts where beliefs might be socially scrutinized. This suggests people regulate expressed beliefs partly for social acceptance, though the effect sizes are modest.
- Impression formation and minimal continued influence. When misinformation is about a person's character (e.g., "this person committed domestic violence") rather than an event, corrections are surprisingly effective. Retracting negative behavioral information about a person produces near-complete recovery of impressions, unlike event-based scenarios. Ecker proposes this may reflect cognitive differences: retracting a cause leaves an event's explanation incomplete and cognitively threatening (demanding a replacement explanation), whereas retracting a behavior about a person leaves the mental model of that person largely intact.
Notes
A comprehensive seminar presentation from a leading experimentalist in the psychology of misinformation. Ecker's core contribution is systematic empirical study of the social factors shaping continued influence—a departure from earlier work focused on cognitive mechanisms alone. The replication failures around political asymmetry and backfire effects are particularly important for the field, as they overturn widely cited findings and caution against designing interventions based on assumed asymmetries. The work on person impressions versus event reasoning hints at deeper cognitive distinctions that deserve further investigation. Strong emphasis on experimental control, large samples (especially in US replication), and honest reporting of null results. Open about methodological tensions (e.g., risk of retrofitting belief ratings to inference responses) and limitations of lab-based inferences about real-world misinformation.