People often have a preference for "self-verifying" feedback that confirms their existing self-views. Self-verification can reinforce existing self-views and prevent opportunities to learn from alternative perspectives, as when people with low self- esteem prefer feedback that validates negative self-beliefs. Past work suggests that a major driver of self-verification is a desire for accurate self-assessment, where disconfirmatory feedback that contradicts existing self-views creates doubt about the credibility of the feedback source. The aim of this study was to develop a formal account of self-verification based on a Bayesian model of source reliability. Findings from a behavioral experiment aligned with the model’s prediction that confirmatory feedback about traits central to one’s self-concept enhances the perceived reliability of a source, while disconfirmatory feedback leads to lower reliability and disinterest in further feedback. This approach clarifies why seemingly biased feedback seeking behaviors may be motivated by rational epistemic concerns about source credibility.