|Abstract||Although previous research has shown that mere thought can polarize attitudes, no previous studies have examined whether a specific type of thought, namely extrapolating from known to unknown attributes of an attitude object, can cause attitudes toward that attitude object to become more extreme or polarized. Two studies tested the predicted relationship between extrapolation and polarization, specifically testing the prediction that extrapolation can cause people who know a few moderately negative attributes of a group to adopt even more negative attitudes toward that group. Study 1 found that this polarizing effect of extrapolation is mediated by source monitoring errors. Study 2 found that the effect is also mediated by perceived valence of cognitive associations to the attitude object. The results add to understanding of how source monitoring and biased evaluation processes can contribute to self-radicalization.