Three needs, and what the stories deliver.
The idea
After a frightening event, 'someone planned this' can feel steadier than 'nobody knows why'. That's the shape of a conspiracy theory: a story that puts an event down to powerful people plotting in secret. Conspiracy theories can draw people in by promising to meet real needs. One is wanting to understand: a huge, confusing event can seem to call for a huge, hidden cause. Another is wanting safety and a sense of control. A third is belonging, and thinking well of yourself and your group. Many of these stories are also hard to prove wrong. They claim the plotters cover their tracks, so anyone arguing against the story can look like part of the plot.
In US surveys, belief in at least one of these theories was common. So this pull isn't something that only happens to other people. The patterns researchers found describe averages; they are not labels for anyone. Feeling excluded, or wanting to stand out as unique, has been linked to stronger belief. So has feeling anxious or powerless, though it's not settled that these feelings cause the belief. Suspicion can also come from experience. Some communities learn to mistrust those in power because those particular people couldn't be trusted. If an official body once misled someone's family, that person may meet the next official statement with more suspicion.
Studies so far suggest these theories often leave people's needs less met, not more. Reading them tends to lower people's sense of control. In tests, they also lowered trust in government bodies, even ones the story wasn't about. And a story can change your view while you hardly notice the change. Most of these tests, though, used students and people signed up to answer surveys. They weren't especially disadvantaged or under threat. For people already on the edges of society, the theories may offer some comfort.
In studies, most one-off efforts to change these beliefs shifted them only slightly or not at all. Warnings given before someone meets a theory, sometimes called prebunking, did better, though only short-term effects were measured. A months-long university course on telling real science from things that only look scientific also did well. In a large study, patient chatbot conversations built around the person's own evidence brought belief down. It was still lower two months later. That study didn't test the same talk between friends or family. The journal later raised a concern about the study and published a correction. The concern is now listed as withdrawn, and the authors' corrected figures showed much the same fall.
Want the research, names and dates? They are in Go deeper, at the end.
Any direction can feel better than none.
Picture it
Picture being lost in thick fog when a sign appears saying 'This way'. Any direction can feel better than standing still. Whether the sign leads home is a separate question.
For a child: A kid whispers to two friends, 'The teachers keep the nicest desserts for themselves.' Now the three of them have a secret club, and it feels great. A shared secret can feel good, whether or not it's true.
Five steps drawn from the findings.
How it applies
One way a conversation might go draws on these findings, though the steps are our own, untested idea. Asking what drew them in can show which need the story speaks to for them. Asking for their strongest evidence narrows the talk to one thing you can check. Answering that one point, clearly and with evidence, keeps the talk from turning into a pile of facts. That is close to what lowered belief in the chatbot study above. Mocking and general counter-arguments did little in other studies. Agreeing where they're right is honest. Guidance from a public body in Europe, not a study, recommends detailed questions about the theory and no ridicule. Leaving time, and returning to it another day, is in that guidance too. None of the approaches studied changed every mind.
Dinner tables, group chats and late-night forums.
Where you’ll see it
- At a family meal: someone says a big blackout was planned, and the table either laughs or goes quiet.
- In a group chat: a clip promises to show "what they don't want you to know".
- Online, after a lonely stretch: a forum where everyone seems to understand you and share a secret.
Signs it’s happening
- Arguments against the theory get read as more proof of the plot.
- It turns a big, frightening event into a story with a clear villain.
- Believing it comes with company, and a sense of being let in on a secret.
Same video, two very different replies.
Before and after
A friend sends you a video and says it proves a cover-up.
"How can you fall for this? It's ridiculous." The talk ends before their reason is heard.
"Which part convinced you most?" Asking which part convinced them turns the talk to one piece of evidence you can look at together.
A real plot, on the record.
A story from history: Watergate, 1972 to 1974
- The Watergate break-in took place on June 17, 1972.
- To look into alleged offences arising from it, a special prosecutor's office was created.
- A grand jury, in March 1974, charged seven White House aides with perjury and obstruction of justice. A grand jury is a panel of citizens that decides on charges. Perjury is lying under oath; obstruction of justice is getting in the way of an investigation or a trial.
- Richard Nixon, the president, was named an 'unindicted coconspirator': named as part of the conspiracy but not charged.
- On August 9, 1974, he resigned.
Real conspiracies have happened, and this one is on the record. 'Have real conspiracies happened?' and 'Is this one real?' are different questions. What answers the second is the evidence for that one.
Quick check3 questions · nothing is saved
1. What do studies so far suggest these theories often do to the needs they promise to meet?
2. In the studies, what did mocking believers do to their beliefs?
3. In the large conversation study, who talked with the people who believed a theory?
One question worth keeping.
In one line
A theory's pull often comes from the needs it promises to meet. Whether it's true is a question of evidence.
Is there a belief you've argued against without ever asking what drew someone to it?
Go deeperThe research, names and dates
Where this comes from
In a 2017 review, Karen Douglas, Robbie Sutton and Aleksandra Cichocka sort these theories' appeal into three motives. Their labels are epistemic, for making sense of your world; existential, for feeling safe and in control; and social. Social means wanting to think well of yourself and your group. They report that belief seems stronger when events are big or significant and leave people unsatisfied with 'mundane, small-scale explanations'.
Across four national US surveys run from 2006 to 2011, 'half of the American public' consistently backed at least one conspiracy theory. A 2023 review in Nature Reviews Psychology reports higher belief in people who feel isolated, lonely or rejected. It reports somewhat higher belief, too, in people with a strong need to feel unique. Early experiments tied feeling out of control to belief, but repeat studies gave mixed results. The review says this mixed evidence casts doubt on the idea that a lack of control causes conspiracy beliefs. It adds that communities sometimes come to mistrust elites because those elites are not trustworthy.
According to the 2017 review, research so far gives no sign that believing these theories fulfils people's motivations. In experiments, reading such theories cut trust in government institutions, even ones the theories were not about. A 2008 study had students read theories about Princess Diana's death. They underestimated how much their own attitudes had been influenced. The 2017 review's authors warn that most experiments used students and survey panelists. Those people were not especially disadvantaged or under threat.
A 2023 review covered 25 studies with 7,179 people. Most of the efforts it looked at left conspiracy beliefs unchanged. All five inoculation, or prebunking, efforts it counted lowered belief, but only short-term effects were measured. A single three-month university course on telling science from pseudoscience had a large effect. In a 2024 Science study, believers talked their theory through with an AI chatbot, in personalized conversations built on evidence. Belief dropped by about 20%, and the drop was still there two months later. The method was tried with a chatbot, not between friends or family. Science issued an expression of concern in June 2026, and a correction on 22 October 2026. The authors' correction materials put 2,044 people in the main analysis, where 2,190 were first reported. Their figures still show belief falling about 21.9% in one study and 19.7% in the other. Between just after the conversation and two months later, they found no significant change. The publisher's records now title the concern 'RESCINDED', and we found no readable notice saying why. So the corrected study still reports a fall in belief, still lower two months later.
What moved these beliefs more vs less, in studies
| Compared | Did better | Did little |
|---|---|---|
| When | A warning before someone meets the theory | Counter-arguments after they have met it |
| How | A patient chatbot talk about the person's own evidence | Ridiculing the people who believe it |
| Scope | One months-long course on telling science from pseudoscience | Most single, one-off attempts |
Sources are a 2023 review (25 studies, 7,179 people) and a 2024 study of chatbot conversations. The warning studies tracked short-term effects only.
More examples
Some everyday moments, and an opening that fits the steps above.
- If a coworker says a big company secretly planned its rival's collapse, 'What first made you think so?' may open more than 'That's ridiculous.'
- If a friend sends a video that explains everything, asking which moment convinced them keeps the talk to one thing.
- If a neighbour calls an official report a cover-up, 'What's the evidence for this particular one?' keeps the door open.
What if…?
- What if correcting them makes them believe it more?
- Researchers call that a backfire effect, where a correction pushes someone further into the false belief. A 2020 review called backfire effects "not a robust empirical phenomenon". The same review notes a debate in the field about whether they happen at all. A careful correction probably won't make matters worse, though by itself it may not shift much.
- What if gentle doesn't work, and mockery does?
- In 2016 an online experiment in Hungary tested 813 people. Both reasoned arguments and ridicule lowered their belief in conspiracy theories. A later review of 25 studies judged ridicule's effect to be very small. That experiment ran online; it wasn't a chat with someone you know. The European Commission, the EU's executive, says "Don't ridicule" in its guidance on these talks. That is a public body's advice rather than a research finding.
- What if the theory turns out to be true?
- Researchers in this field note that history shows business and political elites, again and again, conspiring "against public interests". In 1996, one US company agreed to plead guilty over its part in two international conspiracies to fix prices. It agreed to a $100 million criminal fine.
Words you may meet
- conspiracy theory
- A belief or explanation that says an event or situation is the result of a secret plan made by powerful people. The label describes the shape of the claim; it does not by itself decide whether the claim is true. Glossary →
- backfire effect
- When a correction makes someone believe the false claim it was correcting more, not less. Glossary →