<p>Adding warning labels to clickbait is a common practice to help users recognize the risks associated with online information. However, the effectiveness of these labels remains controversial. This study aims to investigate the effectiveness of different designs of warning labels, including textual and image information, in reducing users' reading behavior of clickbait. Thirty-six adults participated in an experiment involving twelve conditions: four typical warning label types (percentage label, star label, traffic light label, and traffic light label with reference values) × three additional text variations (no text, concluding text, and descriptive text). Participants' reading behavior change rates, associated feedback-related negativity (FRN), and FRN amplitude differences (FRN _Non-winning – FRN _Winning) were measured. The warning label type had a significant effect on the behavior change rate and FRN amplitude and traffic light labels with reference values were significantly more effective than percentage labels and star labels. In contrast, the additional text type had no significant effect on either the behavior change rate or the FRN amplitude. In addition, only in the traffic light labels with reference values condition, there is a significant correlation between the FRN amplitude difference and the reading behavior change rate. Traffic light labels with reference values (with up to 66% reading behavior change rate) should be prioritized for use. Adding additional textual information to a warning label may not reduce the user's reading behavior of the clickbait.</p>