How Recommendation Algorithms and Highly Liked Comments Influence Cognitive Bias on Social Media

Abstract

Social media platforms in the modern digital world mainly depend on user interactions and recommendation algorithms to determine how much content is seen. Highly liked comments are one type of interaction that might be crucial in promoting users' cognitive biases. This study examines how recommender systems and social mechanisms work together to amplify biased thinking and weaken information diversity. This study employs a mixed-method approach, combining survey responses from young users with web-scraped comment data from Chinese platforms Douyin. The analysis focused on patterns in top-rated comments, their emotional tone, and how often they reflected polarized or repetitive viewpoints. Survey data was used to understand how users perceive the impact of algorithms and whether they can recognize bias in content they see on a daily basis. Findings suggest that highly endorsed comments often reflect simplified, emotionally charged opinions that align with existing beliefs. Algorithms are more likely to encourage these comments, which feeds a vicious cycle of biassed content creation and dissemination. Furthermore, users frequently accept this information without challenging its objectivity, which over time exacerbates the cognitive imbalance. This study intends to highlight the need for more transparent platform design and higher media literacy. By understanding how digital systems shape thinking, users and educators can better navigate online space, because what is popular is not always the truth, and more reasonable efforts are needed to create a safe and barrier-free online world for users.