HCL+PrioR mtg. Greta Sanna. Monday 9/29 1:30-2:30pm ET
10 views
Skip to first unread message
Antonio Alonso Arechar
Sep 23, 2025, 2:16:33 PM9/23/25
to Human Cooperation Lab, Sanna, Greta
Good afternoon everyone,
Greta Sanna will present next week (9/29 1:30-2:30pm ET; see blurb below).
Lab meeting will be hybrid, with a zoom link (https://mit.zoom.us/j/91655448125) and in person in the conference room on the 15th floor of E94 (1579; 245 First Street).
Best regards,
Antonio
---
Reasonably partisan: Modelling Belief Updating in Politically Charged Contexts
How rational are people when updating beliefs in politically charged environments? In this talk, I present findings from a study with 200 US participants examining belief updating in response to corrections of politically relevant misinformation. Using a combination of behavioural data and Bayesian computational modelling, we find that participants’ belief updates closely match rational Bayesian predictions, even in the face of identity-congruent or identity-incongruent information. Contrary to predictions from the Continued Influence Effect, participants reliably updated their beliefs when provided with corrective evidence, returning to baseline levels regardless of political alignment, worldview congruence, or extremism. These results suggest that apparent resistance to belief updating reflects the strength of specific beliefs rather than intrinsic identity-based bias. Notably, even self-identified extremists were responsive to sufficiently strong evidence. Our findings provide insights into the patterns of belief updating in politicised contexts and demonstrate how computational modelling can illuminate the processes underlying rational and seemingly irrational belief updating.
How rational are people when updating beliefs in politically charged environments? In this talk, I present findings from a study with 200 US participants examining belief updating in response to corrections of politically relevant misinformation. Using a combination of behavioural data and Bayesian computational modelling, we find that participants’ belief updates closely match rational Bayesian predictions, even in the face of identity-congruent or identity-incongruent information. Contrary to predictions from the Continued Influence Effect, participants reliably updated their beliefs when provided with corrective evidence, returning to baseline levels regardless of political alignment, worldview congruence, or extremism. These results suggest that apparent resistance to belief updating reflects the strength of specific beliefs rather than intrinsic identity-based bias. Notably, even self-identified extremists were responsive to sufficiently strong evidence. Our findings provide insights into the patterns of belief updating in politicised contexts and demonstrate how computational modelling can illuminate the processes underlying rational and seemingly irrational belief updating.
Reply all
Reply to author
Forward
0 new messages