The Illusory Truth Effect: Repetition as Evidence

Lynn Hasher, David Goldstein and Thomas Toppino had students rate the truth of sixty plausible statements — some true, some false, all obscure enough that most people would not know. Then they brought them back two weeks later and did it again, and again two weeks after that. A third of the statements repeated across sessions. The rest were new each time.

Repeated statements were rated as more probably true. It made no difference whether they were true.

That was 1977, and the illusory truth effect has since become one of the most robust findings in the psychology of belief. It has been replicated with news headlines, with advertising claims, with statements about consumer products, and — in the most uncomfortable line of work — with statements the participants demonstrably knew to be false.

That last finding, from Lisa Fazio and colleagues in 2015, is the one that changes what the effect means. Prior knowledge does not protect you. Participants who could correctly answer a question when asked directly still rated a repeated false statement about the same fact as more true than an unrepeated one. Knowledge and fluency compete at the moment of judgment, and fluency does not lose as often as it should.

The mechanism is the same one as the previous chapter, applied to a different output. A repeated statement is easier to process on the second encounter, and processing ease is one of the cues the mind uses to assess truth — reasonably enough, since in ordinary life we tend to encounter true things more often than false ones, and things we already know are easy to process. The cue is a decent heuristic in an environment where repetition is a byproduct of truth. It becomes a catastrophic one in an environment where repetition can be purchased.

Which is the environment we now live in, and the reason this chapter sits underneath the propaganda material in Part 3 and the disinformation material in Part 8. Volume substitutes for evidence, not by convincing anyone in particular but by making a claim progressively easier to think. The claim does not have to be argued. It has to be present.

Several practical consequences follow, and some of them are counterintuitive.

Repeating a false claim in order to debunk it exposes people to the claim again, and there is evidence that corrections can increase familiarity with the falsehood over time even when they succeed in the short term. Effective corrections lead with the true statement, state the false claim only once, explain why it is wrong, and end on the truth — a structure the misinformation literature calls a truth sandwich.

Prebunking, covered in chapter 328, works better than debunking for the same reason: it gets the fluency advantage on the right side.

And personally: the third time a claim feels obviously true, having felt uncertain the first time, is exactly when to check where you heard it. The feeling of having always known something is not evidence of having verified it. Very often it is a record of how many times you have been exposed to it, and that number is something other people can set.