With Dr Jennifer Allen and Reed Orchinik
Concerns about false information masquerading as news are widespread in part because falsehoods are seen as an effective way to mislead and inflame. However, other studies have shown that factually accurate but biased information can also affect beliefs and attitudes. Do falsehoods influence opinions more than biased but factual information? We address this question by conducting a randomized experiment that uses large language models (LLMs) to synthetically generate articles that systematically vary in frame (positive vs. negative) and veracity (accurate vs. exaggerated vs. fabricated). We find that while lying does have a large effect on downstream feelings and opinions, it is no larger than the effect of biased presentation of accurate facts. Our findings challenge the notion that falsehoods are necessary to manipulate public opinion and highlight the potential role of practices like negativity bias and selective emphasis in shaping public misperceptions.
Dr Jennifer Allen is a postdoctoral researcher in the Computational Social Science Lab at the University of Pennsylvania. In Fall 2025, she will join NYU Stern as an Assistant Professor of Technology, Operations, and Statistics and as a core faculty member of the Center for Social Media and Politics. She received her PhD in Management Science from MIT in 2025 and previously worked at Meta and Microsoft Research. Her research interests include misinformation, political persuasion, and the wisdom of crowds.
We process information in environments that are increasingly complex and attentionally demanding – ones that are often argued to exploit our cognitive biases. Conventional wisdom suggests that biases such as the truth bias (the tendency to assume information is true) and the illusory truth effect (whereby repetition increases perceived truthfulness) are immutable flaws in human reasoning. However, using large online experiments and computational models, I demonstrate that these biases are not rigid but adaptive – tailored to the statistical structure of the environment. In low-credibility environments, the truth bias and the illusory truth effect diminish or even reverse. Rather than being inherently flawed, these phenomena appear to reflect adaptively rational inference processes that aid us in our daily lives (where most information and sources are credible), but flexibly recalibrate based on the quality of the information landscape.
Reed Orchinik is a PhD candidate in Management Science at MIT Sloan. His research focuses on how information environments shape cognition including the ways we process information like news, persuasive messages, and marketing communications. With the use of computational models, experiments, and large-scale data, he links cognitive processes with the environments that they operate in to better understand the (sometimes surprisingly rational) foundations for important phenomena like polarization.
postdoctoral researcher in the Computational Social Science Lab, University of Pennsylvania
Jennifer received a PhD in Management Science from MIT & previously worked at Meta and Microsoft Research. Her research interests include misinformation, political persuasion, and the wisdom of crowds
PhD candidate in Management Science , MIT Sloan
Reed's research focuses on how information environments shape cognition including the ways we process information like news, persuasive messages, and marketing communications.