Evidence resistance in polarized information environments: updating political beliefs and trust in news sources

Misinformed and polarized beliefs often persist despite repeated fact-checking efforts. This resistance to evidence is amplified by contextual features of political information environments and underlies political motives that distort judgments about which policy information is accurate and which media source is reliable. Across two preregistered experiments (N = 735), participants evaluated the accuracy of political news headlines from a reliable and an unreliable source. After each initial evaluation, fact-checking feedback was provided followed by a second evaluation. We modelled inference using a dual reinforcement learning framework which captured within-trial belief updating and between-trial updating of trust in sources. We assessed the influence of four contextual factors: politically charged content, partisan source cues, political congruence of fact-checking evidence, and the type of news engagement (accuracy assessment vs news sharing). Results revealed that those who tended to incorporate fact-checking evidence into their beliefs were also likely to update their trust in sources. Despite this positive link, belief updating and trust updating were influenced by different factors. Belief updating followed near Bayesian dynamics and was compromised by politically charged content but improved when individuals engaged in news sharing. In contrast, trust updating was more resistant to fact-checking and was undermined by partisan source cues, politically-incongruent fact-checks, and news sharing. These findings suggest that different features of political information environments may enhance misinformation susceptibility and belief polarization through distinct effects on belief updating and trust in news sources. Recognizing these divergent pathways can inform targeted interventions to curb political misinformation and belief polarization.