Does deliberation decrease belief in conspiracies?

Contributions

• Experimental test of whether deliberation (via two-response paradigm with time pressure and cognitive load) causally reduces conspiracy belief, addressing limitations of prior correlational evidence
• Comparison of two competing mechanisms: deliberation-accuracy (deliberation should decrease false belief) vs. deliberation-coherence (deliberation increases coherence with prior beliefs, so effects depend on mindset)
• Evidence that deliberation's effectiveness is moderated by prior conspiracy beliefs—strongest for people at extremes, minimal for intermediate believers
• Finding that refutation arguments show no deliberation effect, while conspiracy explanations do

Method

Two large online experiments using the two-response paradigm:

Paradigm: Participants give intuitive responses to conspiracy theories under time pressure and cognitive load (dot-matrix task), then give deliberative responses without constraints.

Study 1 (N=1,028): Well-known conspiracies (moon landing, Princess Diana, 9/11, JFK, water fluoridation, chemtrails). Participants rated agreement with conspiracist explanations, official explanations, and refutation arguments.

Study 2 (N=1,000): Lesser-known conspiracies (Deepwater Horizon, Korean Airways Flight 007, Marilyn Monroe, Jonestown). Different presentation format to maximize time pressure. Measured both truth discernment and accuracy judgments.

Conspiracy mindset: Measured via two items on general conspiratorial thinking ("small groups secretly manipulate world events," "important information deliberately concealed from public").

Analysis: Linear mixed-effect models with z-scored agreement as outcome, deliberation condition, explanation type, and conspiracy mindset as fixed effects; random intercepts and slopes by participant and item.

Results

Study 1 (well-known conspiracies):

• No main effect of deliberation on truth discernment (p = .308)
• No three-way interaction with conspiracy mindset (p = .191)
• Conspiracy mindset predicts truth discernment ability (higher mindset = lower accuracy, B = -0.44)

Study 2 (lesser-known conspiracies):

• Deliberation significantly increased truth discernment overall (B = 0.16, p = .003)
• Critical three-way interaction with conspiracy mindset (B = -0.21, p < .001): effect of deliberation depends on prior beliefs
• Post-hoc quartile analysis shows deliberation helps at extremes (bottom quartile: B = 0.38, p = .026; top quartile: B = 0.50, p = .006) but not in middle quartiles
• No deliberation effect for refutation arguments vs. official explanations (p = .170)

Pattern: People in bottom conspiracy-belief quartile decreased agreement with conspiracies after deliberation; people in top quartile decreased belief in conspiracies but did not change belief in official explanations. Middle quartiles showed minimal change.

Notes

Strengths:

• Robust experimental design with strong manipulation of deliberation via time pressure and cognitive load
• Large samples (N=1,000+) with preregistration
• Tests both accuracy-driven and coherence-driven accounts of deliberation
• Clear distinction between types of arguments (conspiracies, official, refutations)

Limitations:

• Null effect in Study 1 suggests familiarity matters; people with settled views about well-known conspiracies may not benefit from deliberation
• Conspiracy mindset measure uses only two items; may not capture specific commitment to particular conspiracies
• "Extreme" believers in the sample were still willing to participate in Mechanical Turk study—may not represent deeply committed real-world conspiracists
• No causal link between prior beliefs and deliberation response (beliefs not experimentally manipulated)
• No effect on refutation arguments; unclear why

Interpretation: The results challenge the simple dual-process narrative that deliberation universally improves accuracy. Instead, they support a coherence-based mechanism where deliberation strengthens whatever beliefs a person enters with. This has practical implications: deliberation-focused interventions may be most effective for people who lack strong prior beliefs (the "apathetic" middle), not for committed believers on either side. The finding that some deliberation effects exist even for high conspiracists is noteworthy but comes with caveats about the representativeness of online research participants.