Mr. Zimmermann, why are false stories so hard to correct?
Florian Zimmermann: In our study, we show that false stories shift people’s beliefs substantially.
When people read or listen to a false story, which contains evaluative or emotional content, they
vividly picture what is happening, a process we call “mental simulation”. It seems that the brain
replicates what is being described in a way similar to a real experience. This changes how people think
about the information they have just learned and persists even after debunking.
False quantitative information, presented as neutral factual claims, does not have such a lasting effect.
According to our findings, people are more confident in beliefs formed after reading false stories
compared to false statistics, even though these beliefs are further away from the rational benchmark.
We also observed that at least some people seem completely unaware that they are influenced by
false stories.
Could you give us a real-life example of a false story that was hard to debunk?
Sure, let us take the allegations of voter fraud in the 2020 U.S. presidential election, for example. They were largely sustained by detailed stories – claims about specific election officials, precincts,
voting machines, and suitcases of ballots – rather than by statistical evidence, which pointed in the
opposite direction. Independent reviews, court rulings, and recounts debunked the specific claims. Yet,
many people in the U.S. continued to believe the election had been compromised.
If official debunking does not work, how should falsehoods and deepfakes be corrected?
Our findings suggest that quantitative rebuttals are not enough to correct story-based falsehoods,
because they work via a different cognitive “channel”. Effective corrections may need to match the
format of the falsehood they target – mental simulation. According to our research, counter-stories
that engage the same imaginative and emotional channels may be more effective than dispassionate
fact-checks. They could be part of corrective interventions. With the rise of AI-generated content, such
as photos, videos, or audio files, which make it easy to package falsehoods as vivid, emotionally
compelling stories, we think there is an urgent need to find effective antidotes.