When Linguistic Uncertainty Spreads Across Pieces of Information: Remembering Facts on the News as Speculation

Modern media enable rapid reporting that does not refer to facts alone but is often interspersed with unconfirmed speculations. Whereas previous research has concentrated primarily on how unconfirmed contents might propagate, potential memory effects of reporting confirmed facts among speculations have so far been widely disregarded. Across four experiments, we show that the presence of speculative news (indexed by uncertainty cues such as "might") can reduce the remembered certainty of unrelated facts. The participants read headlines with exclusively speculative news, exclusively factual news, or a mixture of both. Our results indicate that uncertainty cues spread onto one's recollection of unrelated facts after having read a mixture of facts and speculations. This tendency persisted when both types of news were presented sequentially (e.g., factual news first), suggesting that the presence of speculative news does not specifically affect encoding-but can overshadow memories of facts in retrospect. Further, the tendency to misremember facts as speculations emerged even when the proportion of speculations among factual news was low (6/24 headlines) but increased linearly with the number of speculations intermingled. Given the widespread dissemination of speculative news, this bias poses a challenge in effectively getting confirmed information across to readers. Public Significance Statement This study demonstrates that intermingling speculative, unconfirmed news (e.g., "X might have caused Y") can lead readers to misremember contentwise unrelated factual news (e.g., "A caused B") as unconfirmed information. Given that pieces of speculative news are widely spread, this bias poses a challenge in communicating confirmed contents (facts) as such.