When Information Affects Attitudes: The Effectiveness of Targeting Attitude-Relevant Beliefs

31 March 2025, Version 1
This content is an early or alternative research output and has not been peer-reviewed at the time of posting.

Abstract

Do citizens update strongly held beliefs when presented with belief-incongruent information, and does such updating affect downstream attitudes? Though fact-checking studies find that corrections reliably influence beliefs, attitudinal effects are negligible. We argue that such findings may reflect belief relevance - the extent to which specific beliefs bear on attitudes. Using large language models (LLMs), we elicit deeply held issue attitudes and "focal beliefs" that are described as central to those attitudes. We then randomly assign participants to receive either an LLM-generated factual argument targeting (1) their focal belief, (2) an attitude-relevant but unmentioned belief ("distal belief"), or (3) a placebo. In experiments with two large online convenience samples, we show that counterarguments successfully decrease both focal and distal belief strength, with effects persisting after one week. More importantly, focal belief counterarguments produce larger and more durable attitude change than distal counterarguments.

Keywords

persuasion
fact-checking
generative AI
large language models
political psychology
motivated reasoning
Bayesian updating

Supplementary materials

Title
Description
Actions
Title
Appendix
Description
Supplementary appendix
Actions

Comments

Comments are not moderated before they are posted, but they can be removed by the site moderators if they are found to be in contravention of our Commenting Policy [opens in a new tab] - please read this policy before you post. Comments should be used for scholarly discussion of the content in question. You can find more information about how to use the commenting feature here [opens in a new tab] .
This site is protected by reCAPTCHA and the Google Privacy Policy [opens in a new tab] and Terms of Service [opens in a new tab] apply.