Illusory truth effect

The fact everyone has heard

In the true-or-false round of a quiz night, the question is “Bulls are enraged by the color red.” One team member doesn’t hesitate: “True. I’ve heard that my whole life.”

Nothing in that answer is evidence about bulls. It reports how often the saying has come up, not where it came from or whether anyone tested it. (The saying doesn’t hold up: what provokes a bull in the ring is the cape’s movement, not its color.) The sense of certainty came from familiarity.

The slogan from one sponsor

For months, a gardening podcast runs the same ad for a fertilizer: “Doubles your blooms in two weeks.” A listener who has never tried it tells a friend, “That stuff is supposed to be really good. It’s kind of known for doubling blooms.”

Every repetition came from a single source with a reason to say it, the company paying for the ad. Hearing it a hundred times is one claim, repeated. Yet it has started to feel like common knowledge (“it’s kind of known”), which is the shift the effect describes: repetition changes how true a statement feels without changing the evidence behind it.

Knowing better isn’t a shield

A student drills a shared deck of geography flashcards every evening for a week. One card, made by a classmate, wrongly says the Nile empties into the Red Sea. She learned in school that it reaches the Mediterranean. On the quiz, she hesitates, then picks the Red Sea.

This is the less obvious form. It isn’t limited to claims you know nothing about: in experiments, repeated false statements were rated as more likely true than new ones even when participants knew the correct answer (see Evidence). The effect in those studies is usually a nudge in how true something seems rather than a complete change of mind, but on a close call, a nudge can decide the answer.

• Repetition of the same statement: the standard lab version, in which statements seen once earlier are rated as truer than new ones at test.
• Many repetitions: more exposures raise rated truth further, but with diminishing returns; the largest jump comes from the second encounter.
• Repetition against knowledge: repeated statements that contradict what people know can still gain some credibility (the third example above).

• When the repetitions come from independent sources that each had access to the facts. Several people who checked separately and agree give you more evidence than one person, however often you hear each of them.
• When your exposure has come mostly from reliable sources. Most of what people hear repeatedly is true, so familiarity is often a useful cue. Relying on it is reasonable when the familiarity was built by accurate sources such as textbooks, reference works or your own repeated observation.
• When hearing a claim again sends you to check it, and the check holds up. The belief then rests on the check, not on the repetition.

The test: would this still seem true if I had heard it only once, from wherever it first came from?

Three separate warnings

On the way up a mountain trail, a hiker is told by the ranger at the trailhead that the upper footbridge has washed out. A mile later, two hikers coming down say the same, and a note on the trail register repeats it. She turns back at the junction.

She has heard the claim three times, but the repetitions aren’t the same claim echoed: the ranger, the descending hikers and whoever wrote the note each had their own way of knowing. That’s the independent sources condition above. Repetition like this is evidence, because each source could have found the bridge intact.

A capital city that just sounds right

On a quiz, a student sees “Canberra is the capital of Australia.” She can’t remember where she learned it, but it sounds familiar, and she marks it true.

She’s using familiarity as a cue, which is exactly the mechanism behind the effect. But her familiarity was built by atlases, maps and schoolbooks, which are overwhelmingly accurate about capitals. In that kind of environment, “I’ve seen this many times” is a decent guide to “this is true”. That’s the reliable sources condition above. Compare the bulls and the fertilizer, where the familiar version came from sayings and advertising.

The leading explanation is processing fluency: a statement you’ve encountered before is easier to read and understand, and people take that ease as a sign of truth. A related account holds that repetition strengthens and connects the memory references a statement brings to mind, so it fits better with what you already hold, and fit is read as truth. Researchers treat these accounts as closely related, and it isn’t settled how much each contributes.

The cue is useful often enough to persist. Reviewers of this research point out that people have a general tendency to accept what they’re told because most claims they encounter are true, and that ease of processing is usually a fair signal. The effect is the cost of that shortcut when the environment includes repeated falsehoods.

This is why sheer repetition persuades. Repeating a claim until people accept it is the persuasion technique called proof by repetition (argumentum ad nauseam); the illusory truth effect is the bias that tactic relies on. It is also why repetition can masquerade as an appeal to popularity: a claim that has been repeated sounds as if many people must hold it, even when the repetitions all come from one place.

The illusory truth effect is also the well-established finding behind the claimed “familiarity” form of the Backfire effect: the worry that repeating a myth while correcting it would make it feel truer. The two fit together. The illusory truth effect concerns repetition without a clear correction; when a statement is plainly flagged as false at the time people first see it, studies on average find that the boost disappears or reverses (see Evidence).

Several responses have been tested:

• Judge the claim when you first meet it. Studies that asked people to rate how true each statement was on first exposure found weaker illusory truth effects later than studies that had people read the statements for some other purpose.
• Clear labels at first exposure help; mild ones may not. A meta-analysis found that telling people a statement is false when they first see it removed or reversed the effect on average. A “disputed” tag on fabricated news headlines didn’t stop the effect in one set of experiments.
• Knowing the facts and trying harder don’t reliably protect you. Relevant knowledge didn’t prevent the effect in lab studies, and paying people for each accurate judgment didn’t reduce it.
• Count sources, not encounters. When something feels familiar, ask how many separate people actually checked it. This is practical advice rather than a tested intervention.

Status: replicates robustly. The effect has appeared across many independent labs and materials, in preregistered studies, and in a 2026 meta-analysis whose effect survives correction for publication bias. The caveats are about size and study quality, not existence: the average effect is small to moderate, and the literature has few direct replications and no multi-lab study.

• Hasher, Goldstein and Toppino (1977) first reported it. Participants rated plausible statements, some true and some false, in three sessions spaced weeks apart; some statements recurred across sessions. Ratings of the repeated statements rose from session to session, and the increase was similar whether the statements were true or false.
• Dechêne and colleagues (2010) meta-analyzed 51 studies and found a medium effect (d ≈ 0.39 comparing a statement’s ratings across encounters, and d ≈ 0.50 comparing repeated with new statements). A later review noted the meta-analysis did not include formal checks for publication bias.
• Fazio and colleagues (2015) found illusory truth effects even for statements that contradicted participants’ knowledge, measured with normed estimates and a knowledge check at the end of the experiment. Because measuring knowledge afterward could be affected by the experiment itself, Fazio (2020) measured knowledge beforehand and again found that repeated false statements were rated truer than new ones when they contradicted what participants knew.
• Pennycook, Cannon and Rand (2018) found that a single prior exposure increased the perceived accuracy of fabricated news headlines, both in the same session and a week later, including headlines labeled as disputed by fact checkers. Repetition did not raise belief in entirely implausible statements, such as that the earth is a perfect square.
• Hassan and Barber (2021) repeated trivia statements up to 27 times. Perceived truth kept rising with repetitions but logarithmically, with the biggest gain from the second exposure.
• Henderson, Simons and Barr (2021), a preregistered Registered Report with 567 participants analyzed, found the effect immediately and after a day, a week and a month, but shrinking with time: the difference between repeated and new statements fell from 0.67 to 0.14 points on a 7-point scale.
• Speckmann and Unkelbach (2022), in a preregistered, high-powered experiment, found that paying for every truth judgment did not reduce the effect.
• Henderson, Westwood and Simons (2022), a preregistered systematic map of 181 studies, raised concerns about the literature rather than the effect: every study was run at a Western university, most repeated trivia statements within a single session, only 3% of papers included an external replication, and 96% of studies that reported a test found a positive result, a sign of possible publication bias. They called for multi-lab studies.
• Ye and colleagues (2026) meta-analyzed 182 studies (366 effect sizes, 31,184 participants, 1977–2025). They found clear signs of publication bias and, after correcting for it, a small but reliable effect (g = 0.37, 95% CI 0.30 to 0.44), with large differences between studies. The effect was similar for true and false statements, smaller for news headlines than for standard statements, weaker when people judged truth at first exposure, and reversed on average when statements were flagged as false at first exposure. They caution that their correction method is less accurate when studies differ this much.

Is it really an error? Some researchers argue that relying on repetition is sensible in a mostly accurate information environment. A 2025 preprint (Orchinik, Rand and Bhui, not yet peer reviewed) reports a preregistered experiment with 4,947 participants in which the effect was substantially smaller when most statements people saw were false, and absent for most participants in that condition. This bears on why the effect exists and when it misleads, not on whether it occurs.

What remains uncertain. How much the lab effect, typically measured with trivia and short delays, changes real beliefs over months of repeated exposure; how it varies outside Western student and online samples; and how the explanations based on fluency and memory references relate to each other.